Analysis.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "d90587c5",
   "metadata": {},
   "outputs": [
    {
     "ename": "ModuleNotFoundError",
     "evalue": "No module named 'scipy'",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mModuleNotFoundError\u001b[0m                       Traceback (most recent call last)",
      "\u001b[1;32mc:\\Users\\Max\\My Drive\\Github\\limbic-NHSDigital-RecoveryRates\\Analysis.ipynb Cell 1\u001b[0m in \u001b[0;36m<cell line: 8>\u001b[1;34m()\u001b[0m\n\u001b[0;32m      <a href='vscode-notebook-cell:/c%3A/Users/Max/My%20Drive/Github/limbic-NHSDigital-RecoveryRates/Analysis.ipynb#W0sZmlsZQ%3D%3D?line=5'>6</a>\u001b[0m \u001b[39mfrom\u001b[39;00m \u001b[39mdatetime\u001b[39;00m \u001b[39mimport\u001b[39;00m date\n\u001b[0;32m      <a href='vscode-notebook-cell:/c%3A/Users/Max/My%20Drive/Github/limbic-NHSDigital-RecoveryRates/Analysis.ipynb#W0sZmlsZQ%3D%3D?line=6'>7</a>\u001b[0m \u001b[39mimport\u001b[39;00m \u001b[39mmath\u001b[39;00m\n\u001b[1;32m----> <a href='vscode-notebook-cell:/c%3A/Users/Max/My%20Drive/Github/limbic-NHSDigital-RecoveryRates/Analysis.ipynb#W0sZmlsZQ%3D%3D?line=7'>8</a>\u001b[0m \u001b[39mfrom\u001b[39;00m \u001b[39mscipy\u001b[39;00m\u001b[39m.\u001b[39;00m\u001b[39mstats\u001b[39;00m \u001b[39mimport\u001b[39;00m chi2_contingency \n\u001b[0;32m      <a href='vscode-notebook-cell:/c%3A/Users/Max/My%20Drive/Github/limbic-NHSDigital-RecoveryRates/Analysis.ipynb#W0sZmlsZQ%3D%3D?line=8'>9</a>\u001b[0m \u001b[39mimport\u001b[39;00m \u001b[39mdatetime\u001b[39;00m\n\u001b[0;32m     <a href='vscode-notebook-cell:/c%3A/Users/Max/My%20Drive/Github/limbic-NHSDigital-RecoveryRates/Analysis.ipynb#W0sZmlsZQ%3D%3D?line=9'>10</a>\u001b[0m \u001b[39mfrom\u001b[39;00m \u001b[39mdateutil\u001b[39;00m\u001b[39m.\u001b[39;00m\u001b[39mrelativedelta\u001b[39;00m \u001b[39mimport\u001b[39;00m relativedelta \u001b[39mas\u001b[39;00m relativedelta\n",
      "\u001b[1;31mModuleNotFoundError\u001b[0m: No module named 'scipy'"
     ]
    }
   ],
   "source": [
    "import requests\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from dotenv import load_dotenv\n",
    "import os\n",
    "from datetime import date\n",
    "import math\n",
    "from scipy.stats import chi2_contingency \n",
    "import datetime\n",
    "from dateutil.relativedelta import relativedelta as relativedelta\n",
    "from iteration_utilities import deepflatten\n",
    "import matplotlib.pyplot as plt\n",
    "from scipy.spatial.distance import cdist\n",
    "import io\n",
    "import sys\n",
    "import scipy\n",
    "from numpy.linalg import norm\n",
    "from scipy.signal import savgol_filter\n",
    "import statsmodels.api as sm\n",
    "import scipy\n",
    "\n",
    "data_folder=\"./data/\"\n",
    "folder_figures_save=\"G:/Shared drives/05_Research/Publications (research papers)\\Clinical/NHS_Digital_PrePost_Limbic/Figures/\"\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9573e01c",
   "metadata": {},
   "source": [
    "## Load data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9810e4f7",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_old=pd.read_csv(data_folder+'NHS_England_Feb-2021.csv')\n",
    "data_new=pd.read_csv(data_folder+'NHS_England_Feb-2022.csv')\n",
    "data_new2=pd.read_csv(data_folder+'NHS_England_May-2022.csv')\n",
    "\n",
    "data_all=pd.concat([data_new, data_old])\n",
    "data_all=pd.concat([data_all, data_new2])\n",
    "data_all.drop_duplicates(inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d7266473",
   "metadata": {},
   "source": [
    "## Define functions for processing data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "eaca3d59",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_recovery_sum(data_all, dates_previous, measure): \n",
    "    recovery=[]\n",
    "    non_recovery=[]\n",
    "    recovery_timeline=[]\n",
    "    non_recovery_timeline=[]\n",
    "    for s in data_all['ORG_NAME2'].unique(): \n",
    "        data_insight_all=data_all[data_all['ORG_NAME2']==s]\n",
    "        r=[]\n",
    "        non_r=[]\n",
    "        for d in dates_previous: \n",
    "            data_months=data_insight_all[data_insight_all['REPORTING_PERIOD_START']==d]\n",
    "            try: \n",
    "                ended_treatment=int(np.array(data_months['MEASURE_VALUE'][data_months['MEASURE_NAME']=='Count_FinishedCourseTreatment'])[0])\n",
    "                percent_recovery=float(np.array(data_months['MEASURE_VALUE'][data_months['MEASURE_NAME']==measure])[0])/100\n",
    "                \n",
    "                recovered=np.round(ended_treatment*percent_recovery)\n",
    "                not_recovered=np.round(ended_treatment*(1-percent_recovery))\n",
    "                recovery.append(recovered)\n",
    "                r.append(recovered)\n",
    "                non_recovery.append(not_recovered)\n",
    "                non_r.append(not_recovered)\n",
    "            except: \n",
    "                recovery.append(0)\n",
    "                non_recovery.append(0)\n",
    "                r.append(0)\n",
    "                non_r.append(0)\n",
    "        recovery_timeline.append(r)\n",
    "        non_recovery_timeline.append(non_r)\n",
    "        \n",
    "    return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
    "\n",
    "\n",
    "\n",
    "def get_recovery_rates(dates_previous, dates_after, measure, data): \n",
    "    recovery_previous, not_recovered_previous, average_number_referrals_before, timeline_recovered_before=get_recovery_sum(data, dates_previous, measure)\n",
    "    if recovery_previous!=0: \n",
    "        recovery_after, not_recovered_after, average_number_referrals_after, timeline_recovered_after=get_recovery_sum(data, dates_after, measure)\n",
    "    else: \n",
    "        return 0, 0, 0, 0, np.array([0,0,0,0]), 0, 0, [], []\n",
    "    try: \n",
    "        p=test_for_significance(recovery_previous, not_recovered_previous, recovery_after, not_recovered_after)\n",
    "        recovery_rate_before=recovery_previous/(recovery_previous+not_recovered_previous)\n",
    "        recovery_rate_after=recovery_after/(recovery_after+not_recovered_after)\n",
    "        recover_split=np.array([recovery_previous,not_recovered_previous,recovery_after,not_recovered_after])\n",
    "        all_observation=recover_split.sum()\n",
    "        return recovery_rate_before, recovery_rate_after, p, all_observation, recover_split, average_number_referrals_before, average_number_referrals_after, timeline_recovered_before, timeline_recovered_after\n",
    "    except: \n",
    "        return 0, 0, 0, 0, np.array([0,0,0,0]), 0, 0, [], []\n",
    "\n",
    "def test_for_significance(recovery_previous, not_recovered_previous, recovery_after, not_recovered_after): \n",
    "    before = [recovery_previous, not_recovered_previous] \n",
    "    after = [recovery_after, not_recovered_after] \n",
    "    table=[before, after]\n",
    "    chi2, p, dof, ex = chi2_contingency(table) \n",
    "    return p\n",
    "\n",
    "def get_referrall_rates(dates_previous, dates_after, data): \n",
    "    referrals_before, referrals_perMonth_before=get_referral_sum(data, dates_previous)\n",
    "    if referrals_before!=0: \n",
    "        referrals_after, referrals_perMonth_after=get_referral_sum(data, dates_after)\n",
    "    else: \n",
    "        return 0, 0, 0, 0\n",
    "    return referrals_before, referrals_after, referrals_perMonth_before, referrals_perMonth_after\n",
    "\n",
    "def get_referral_sum(data_all, dates_previous): \n",
    "    referrals=[]\n",
    "    for s in data_all['ORG_NAME2'].unique(): \n",
    "        data_insight_all=data_all[data_all['ORG_NAME2']==s]\n",
    "        r=[]\n",
    "        non_r=[]\n",
    "        for d in dates_previous: \n",
    "            data_months=data_insight_all[data_insight_all['REPORTING_PERIOD_START']==d]\n",
    "            try: \n",
    "                arr=np.array(data_months['MEASURE_VALUE'][data_months['MEASURE_NAME']=='Count_ReferralsReceived'])\n",
    "                ref_months=int(arr[0])\n",
    "                referrals.append(ref_months)\n",
    "            except: \n",
    "                referrals.append(0)\n",
    "    return  np.array(referrals).sum(), np.array(referrals).sum()/len(dates_previous)\n",
    "\n",
    "\n",
    "def prepare_output_vector(results_pre_post): \n",
    "    outcome_vector=[]\n",
    "    outcome_vector.append([int(1)]*int(results_pre_post[0]))\n",
    "    outcome_vector.append([int(0)]*int(results_pre_post[1]))\n",
    "    outcome_vector.append([int(1)]*int(results_pre_post[2]))\n",
    "    outcome_vector.append([int(0)]*int(results_pre_post[3]))\n",
    "    return np.array(list(deepflatten(outcome_vector)))\n",
    "\n",
    "\n",
    "def prepare_pre_post(results_pre_post): \n",
    "    input_vector=[]\n",
    "    input_vector.append([0]*int(results_pre_post[0]+results_pre_post[1]))\n",
    "    input_vector.append([1]*int(results_pre_post[2]+results_pre_post[3]))\n",
    "    return np.array(list(deepflatten(input_vector)))\n",
    "\n",
    "def prepare_side_identifier(results_pre_post, ind): \n",
    "    return np.array([ind]*int(results_pre_post.sum()))\n",
    "\n",
    "def prepare_initial_recover(results_pre_post, initial_recovery): \n",
    "    return np.array([initial_recovery]*int(results_pre_post.sum()))\n",
    "\n",
    "\n",
    "def prepare_input_vector(results_pre_post, ind): \n",
    "    input_vector=[]\n",
    "    input_vector.append(prepare_side_identifier(results_pre_post, ind))\n",
    "    input_vector.append(prepare_pre_post(results_pre_post))\n",
    "    return np.array(input_vector)\n",
    "\n",
    "\n",
    "def get_recovery_service(input_data, months=4, measure='Percentage_ReliableRecovery'): \n",
    "    #Percentage_FirstTreatment6WeeksFinishedCourseTreatment\n",
    "    #Count_ReferralsReceived\n",
    "    service_ind=[]\n",
    "    service=input_data['service']\n",
    "    for k in data_all['ORG_NAME2']: \n",
    "        if k.lower().startswith(service): \n",
    "            service_ind.append(True)\n",
    "        else: \n",
    "            service_ind.append(False)\n",
    "    data_all['service_index']=np.array(service_ind)\n",
    "    data_service=data_all[data_all['service_index']==True]\n",
    "    dt_object = datetime.datetime.strptime(input_data['start_date'],  \"%d/%m/%Y\")\n",
    "    dt_object2 = datetime.datetime.strptime(input_data['final_date'],  \"%d/%m/%Y\")\n",
    "    month_Minus1=dt_object  - relativedelta(months=1)\n",
    "    month_Minus2=dt_object  - relativedelta(months=2)\n",
    "    # today = datetime.date.today()\n",
    "    diff=dt_object2-dt_object\n",
    "    date_after=[dt_object  + relativedelta(months=i) for i in range(months, round(diff.days/30))]\n",
    "    dates_previous=[month_Minus2.strftime(\"%d/%m/%Y\").split(' ')[0], month_Minus1.strftime(\"%d/%m/%Y\").split(' ')[0], dt_object.strftime(\"%d/%m/%Y\").split(' ')[0]] #, '01/01/2021', '01/02/2021'\n",
    "    dates_after_string=[d.strftime(\"%d/%m/%Y\").split(' ')[0] for d in date_after]\n",
    "    return get_recovery_rates(dates_previous, dates_after_string, measure, data_service)\n",
    "\n",
    "def get_referral_service(input_data, months=4): \n",
    "    #Percentage_FirstTreatment6WeeksFinishedCourseTreatment\n",
    "    #Count_ReferralsReceived\n",
    "    service_ind=[]\n",
    "    service=input_data['service']\n",
    "    for k in data_all['ORG_NAME2']: \n",
    "        if k.lower().startswith(service): \n",
    "            service_ind.append(True)\n",
    "        else: \n",
    "            service_ind.append(False)\n",
    "    data_all['service_index']=np.array(service_ind)\n",
    "    data_service=data_all[data_all['service_index']==True]\n",
    "    dt_object = datetime.datetime.strptime(input_data['start_date'],  \"%d/%m/%Y\")\n",
    "    dt_object2 = datetime.datetime.strptime(input_data['final_date'],  \"%d/%m/%Y\")\n",
    "    month_Minus1=dt_object  - relativedelta(months=1)\n",
    "    month_Minus2=dt_object  - relativedelta(months=2)\n",
    "    # today = datetime.date.today()\n",
    "    diff=dt_object2-dt_object\n",
    "    date_after=[dt_object  + relativedelta(months=i) for i in range(months, round(diff.days/30))]\n",
    "    print(date_after)\n",
    "    dates_previous=[month_Minus2.strftime(\"%d/%m/%Y\").split(' ')[0], month_Minus1.strftime(\"%d/%m/%Y\").split(' ')[0], dt_object.strftime(\"%d/%m/%Y\").split(' ')[0]] #, '01/01/2021', '01/02/2021'\n",
    "    dates_after_string=[d.strftime(\"%d/%m/%Y\").split(' ')[0] for d in date_after]\n",
    "    return get_referrall_rates(dates_previous, dates_after_string,data_service)\n",
    "\n",
    "def get_timeline(months): \n",
    "    results_pre_post_limbic=[]\n",
    "    results_pre_post_NHS=[]\n",
    "\n",
    "    outcomes=[]\n",
    "    pre_post=[]\n",
    "    service_id=[]\n",
    "    initial_recovery_rates=[]\n",
    "    limbic_used=[]\n",
    "    number_referral_initial=[]\n",
    "    number_referral_final=[]\n",
    "    service_code=[]\n",
    "    recovery_timeline_limbic_pre=[]\n",
    "    recovery_timeline_limbic_post=[]\n",
    "    serviceName_pre_post_NHS=[]\n",
    "\n",
    "    recovery_timeline_NHS_pre=[]\n",
    "    recovery_timeline_NHS_post=[]\n",
    "    results_raw=[]\n",
    "\n",
    "    for i, input_data in enumerate(inputs_services): \n",
    "        output=get_recovery_service(input_data, months=months)\n",
    "        results_pre_post_limbic.append([output[0], output[1]])\n",
    "        outcomes.append(prepare_output_vector(output[4]))\n",
    "        results_raw.append(output[4])\n",
    "        service_code.append([input_data]*int(output[4].sum()))\n",
    "        pre_post.append(prepare_pre_post(output[4]))\n",
    "        service_id.append(prepare_side_identifier(output[4], i))\n",
    "        initial_recovery_rates.append(prepare_side_identifier(output[4], output[0]))\n",
    "        limbic_used.append(prepare_side_identifier(output[4], 1))\n",
    "        number_referral_initial.append(prepare_side_identifier(output[4], int(output[5])))\n",
    "        number_referral_final.append(prepare_side_identifier(output[4], int(output[6])))\n",
    "        recovery_timeline_limbic_pre.append(output[7])\n",
    "        recovery_timeline_limbic_post.append(output[8])\n",
    "        \n",
    "\n",
    "    for i, input_data in enumerate(all_services): \n",
    "        if (input_data.lower().startswith('insight')==False) & (input_data.lower().startswith('trent')==False) & (input_data.lower().startswith('vita health')==False) & (input_data.lower().startswith('surrey iapt')==False)& (input_data.lower().startswith('all providers')==False): \n",
    "            output=get_recovery_service( {'service': input_data.lower(), 'start_date': '01/12/2020', 'final_date': '01/05/2022'}, months=months)\n",
    "            service_code.append([input_data]*int(output[4].sum()))\n",
    "            if output[0]!=0: \n",
    "                results_pre_post_NHS.append([output[0], output[1]])\n",
    "                serviceName_pre_post_NHS.append(input_data)\n",
    "                results_raw.append(output[4])\n",
    "            outcomes.append(prepare_output_vector(output[4]))\n",
    "            pre_post.append(prepare_pre_post(output[4]))\n",
    "            service_id.append(prepare_side_identifier(output[4], i))\n",
    "            initial_recovery_rates.append(prepare_side_identifier(output[4], output[0]))\n",
    "            limbic_used.append(prepare_side_identifier(output[4], 0))\n",
    "            number_referral_initial.append(prepare_side_identifier(output[4], int(output[5])))\n",
    "            number_referral_final.append(prepare_side_identifier(output[4], int(output[6])))\n",
    "            recovery_timeline_NHS_pre.append(output[7])\n",
    "            recovery_timeline_NHS_post.append(output[8])\n",
    "        else: \n",
    "            print(input_data)\n",
    "\n",
    "\n",
    "\n",
    "    outcomes=np.array(list(deepflatten(outcomes)))\n",
    "    pre_post=np.array(list(deepflatten(pre_post)))\n",
    "    service_id=np.array(list(deepflatten(service_id)))\n",
    "    initial_recovery_rates=np.array(list(deepflatten(initial_recovery_rates)))\n",
    "    limbic_used=np.array(list(deepflatten(limbic_used)))\n",
    "\n",
    "    service_code=np.array(list(deepflatten(service_code)))\n",
    "    number_referral_initial=np.array(list(deepflatten(number_referral_initial)))\n",
    "    number_referral_final=np.array(list(deepflatten(number_referral_final)))\n",
    "    results={'outcomes': outcomes, \n",
    "    'pre_post': pre_post, \n",
    "    'service_id': service_id, \n",
    "    'initial_recovery_rates': initial_recovery_rates, \n",
    "    'limbic_used': limbic_used,\n",
    "    'service_code': service_code,\n",
    "    'number_referral_initial': number_referral_initial,\n",
    "    'number_referral_final': number_referral_final,\n",
    "    'recovery_timeline_limbic_pre': recovery_timeline_limbic_pre,\n",
    "    'recovery_timeline_limbic_post': recovery_timeline_limbic_post,\n",
    "    'recovery_timeline_NHS_pre': recovery_timeline_NHS_pre,\n",
    "    'recovery_timeline_NHS_post': recovery_timeline_NHS_post,\n",
    "    'results_pre_post_NHS': results_pre_post_NHS, \n",
    "    'results_pre_post_limbic': results_pre_post_limbic, \n",
    "    'results_raw':results_raw\n",
    "    }\n",
    "    return results"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b8c1cd84",
   "metadata": {},
   "source": [
    "Define services and time during which Limbic was used"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "id": "4973c0c9",
   "metadata": {},
   "outputs": [],
   "source": [
    "all_services=data_all['ORG_NAME2'].unique()\n",
    "\n",
    "input_data_insight={\"service\":'insight',\n",
    "    'start_date':'01/11/2021',\n",
    "    'final_date':'01/06/2022',\n",
    "}\n",
    "\n",
    "input_data_surrey={\"service\":'surrey iapt',\n",
    "    'start_date':'01/07/2021',\n",
    "    'final_date':'01/06/2022',\n",
    "}\n",
    "\n",
    "input_data_trent={\"service\":'trent',\n",
    "    'start_date':'01/07/2021',\n",
    "    'final_date':'01/06/2022',\n",
    "}\n",
    "\n",
    "\n",
    "input_data_vita={\"service\":'vita health',\n",
    "    'start_date':'01/12/2020',\n",
    "    'final_date':'01/06/2022',\n",
    "}\n",
    "\n",
    "inputs_services=[input_data_insight, input_data_trent, input_data_vita, input_data_surrey]#,  input_data_trent, input_data_vita\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f673b1af",
   "metadata": {},
   "source": [
    "## Create Figure 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "id": "660c3e4d",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n"
     ]
    }
   ],
   "source": [
    "results=get_timeline(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 197,
   "id": "5fe7fa80",
   "metadata": {},
   "outputs": [],
   "source": [
    "results_pre_post_limbic=np.array(results['results_pre_post_limbic'])\n",
    "results_pre_post_NHS=np.array(results['results_pre_post_NHS'])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "id": "ab2af38f",
   "metadata": {},
   "outputs": [],
   "source": [
    "results['recovery_timeline_limbic_post'][3]=results['recovery_timeline_limbic_post'][3][np.isnan(results['recovery_timeline_limbic_post'][3])==False]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "id": "99914d41",
   "metadata": {},
   "outputs": [],
   "source": [
    "averages_post=[]\n",
    "for i in range(17): \n",
    "    average_for_time=[]\n",
    "    try: \n",
    "        average_for_time.append(results['recovery_timeline_limbic_post'][0][i])\n",
    "    except: \n",
    "        i\n",
    "    try: \n",
    "        average_for_time.append(results['recovery_timeline_limbic_post'][1][i])\n",
    "    except: \n",
    "        i\n",
    "    try: \n",
    "        average_for_time.append(results['recovery_timeline_limbic_post'][2][i])\n",
    "    except: \n",
    "        i\n",
    "    try: \n",
    "        average_for_time.append(results['recovery_timeline_limbic_post'][3][i])\n",
    "    except: \n",
    "        i\n",
    "    averages_post.append(np.array(average_for_time).mean())\n",
    "averages_post=np.array(averages_post)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "id": "e0eb0f40",
   "metadata": {},
   "outputs": [],
   "source": [
    "averages_post=averages_post[np.isnan(averages_post)==False]\n",
    "yhat = savgol_filter(np.array(results['recovery_timeline_limbic_pre']).mean(axis=0), 3, 1)\n",
    "\n",
    "yhat2 = savgol_filter(np.array(averages_post), 9, 1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "id": "7ab39e46",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(10, 8))\n",
    "\n",
    "plt.plot([-2,-1,0], yhat, linewidth=4, color='grey')\n",
    "plt.plot([f+1 for f in range(len(averages_post))], yhat2, linewidth=4, color='orange')\n",
    "plt.ylim(.45, .51)\n",
    "plt.ylabel('recovery rates %', fontsize=20)\n",
    "plt.xticks([-2, -1, 0,1, 3, 5, 7, 9, 11,13, 15, 17], ['-2', '-1', '','1','3','5', '7', '9', '11', '13', '15', '17'], fontsize=20)\n",
    "plt.yticks([.45, .46, .47,.48, .49, .50, .51], ['45', '46', '47', '48', '49', '50','51'], fontsize=20)\n",
    "\n",
    "plt.xlabel('months', fontsize=20)\n",
    "\n",
    "image_format = 'svg' # e.g .png, .svg, etc.\n",
    "#image_name = folder_figures_save+'timeline_improvement.svg'\n",
    "\n",
    "#fig.savefig(image_name, format=image_format, dpi=1200)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "id": "cc57bf16",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n"
     ]
    }
   ],
   "source": [
    "results=get_timeline(4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "id": "9d8e695c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table class=\"simpletable\">\n",
       "<caption>Logit Regression Results</caption>\n",
       "<tr>\n",
       "  <th>Dep. Variable:</th>        <td>outcome</td>     <th>  No. Observations:  </th>   <td>835393</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Model:</th>                 <td>Logit</td>      <th>  Df Residuals:      </th>   <td>835386</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Method:</th>                 <td>MLE</td>       <th>  Df Model:          </th>   <td>     6</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Date:</th>            <td>Wed, 02 Nov 2022</td> <th>  Pseudo R-squ.:     </th>  <td>0.004854</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Time:</th>                <td>18:02:30</td>     <th>  Log-Likelihood:    </th> <td>-5.7533e+05</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>converged:</th>             <td>True</td>       <th>  LL-Null:           </th> <td>-5.7814e+05</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Covariance Type:</th>     <td>nonrobust</td>    <th>  LLR p-value:       </th>   <td> 0.000</td>   \n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "    <td></td>       <th>coef</th>     <th>std err</th>      <th>z</th>      <th>P>|z|</th>  <th>[0.025</th>    <th>0.975]</th>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>const</th> <td>   -1.4332</td> <td>    0.021</td> <td>  -69.719</td> <td> 0.000</td> <td>   -1.474</td> <td>   -1.393</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>x1</th>    <td>   -0.0366</td> <td>    0.006</td> <td>   -6.132</td> <td> 0.000</td> <td>   -0.048</td> <td>   -0.025</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>x2</th>    <td>   -0.1110</td> <td>    0.018</td> <td>   -6.261</td> <td> 0.000</td> <td>   -0.146</td> <td>   -0.076</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>x3</th>    <td>    2.7342</td> <td>    0.040</td> <td>   68.791</td> <td> 0.000</td> <td>    2.656</td> <td>    2.812</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>x4</th>    <td> 8.932e-05</td> <td> 6.24e-06</td> <td>   14.318</td> <td> 0.000</td> <td> 7.71e-05</td> <td>    0.000</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>x5</th>    <td>   -0.0004</td> <td> 1.16e-05</td> <td>  -31.781</td> <td> 0.000</td> <td>   -0.000</td> <td>   -0.000</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>x6</th>    <td>    0.2199</td> <td>    0.020</td> <td>   11.083</td> <td> 0.000</td> <td>    0.181</td> <td>    0.259</td>\n",
       "</tr>\n",
       "</table>"
      ],
      "text/plain": [
       "<class 'statsmodels.iolib.summary.Summary'>\n",
       "\"\"\"\n",
       "                           Logit Regression Results                           \n",
       "==============================================================================\n",
       "Dep. Variable:                outcome   No. Observations:               835393\n",
       "Model:                          Logit   Df Residuals:                   835386\n",
       "Method:                           MLE   Df Model:                            6\n",
       "Date:                Wed, 02 Nov 2022   Pseudo R-squ.:                0.004854\n",
       "Time:                        18:02:30   Log-Likelihood:            -5.7533e+05\n",
       "converged:                       True   LL-Null:                   -5.7814e+05\n",
       "Covariance Type:            nonrobust   LLR p-value:                     0.000\n",
       "==============================================================================\n",
       "                 coef    std err          z      P>|z|      [0.025      0.975]\n",
       "------------------------------------------------------------------------------\n",
       "const         -1.4332      0.021    -69.719      0.000      -1.474      -1.393\n",
       "x1            -0.0366      0.006     -6.132      0.000      -0.048      -0.025\n",
       "x2            -0.1110      0.018     -6.261      0.000      -0.146      -0.076\n",
       "x3             2.7342      0.040     68.791      0.000       2.656       2.812\n",
       "x4          8.932e-05   6.24e-06     14.318      0.000    7.71e-05       0.000\n",
       "x5            -0.0004   1.16e-05    -31.781      0.000      -0.000      -0.000\n",
       "x6             0.2199      0.020     11.083      0.000       0.181       0.259\n",
       "==============================================================================\n",
       "\"\"\""
      ]
     },
     "execution_count": 203,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data=pd.DataFrame({'pre_post': results['pre_post'], 'initial_recovery':  results['initial_recovery_rates'],'outcome': results['outcomes'], 'limbic':results['limbic_used'] , 'number_referrals_pre':results['number_referral_initial'] , 'number_referrals_post':results['number_referral_final'] })\n",
    "x=np.transpose(np.array([data['pre_post'], data['limbic'], data['initial_recovery'], data['number_referrals_pre'], data['number_referrals_post']-data['number_referrals_pre'], data['pre_post']*data['limbic']]))\n",
    "x=sm.add_constant(x)\n",
    "model = sm.Logit(data['outcome'], x).fit()\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "id": "9bfdc74f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table class=\"simpletable\">\n",
       "<caption>Logit Regression Results</caption>\n",
       "<tr>\n",
       "  <th>Dep. Variable:</th>        <td>outcome</td>     <th>  No. Observations:  </th>   <td>680519</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Model:</th>                 <td>Logit</td>      <th>  Df Residuals:      </th>   <td>680517</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Method:</th>                 <td>MLE</td>       <th>  Df Model:          </th>   <td>     1</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Date:</th>            <td>Wed, 02 Nov 2022</td> <th>  Pseudo R-squ.:     </th>  <td>4.288e-05</td> \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Time:</th>                <td>18:02:32</td>     <th>  Log-Likelihood:    </th> <td>-4.7080e+05</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>converged:</th>             <td>True</td>       <th>  LL-Null:           </th> <td>-4.7082e+05</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Covariance Type:</th>     <td>nonrobust</td>    <th>  LLR p-value:       </th>  <td>2.092e-10</td> \n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "    <td></td>       <th>coef</th>     <th>std err</th>      <th>z</th>      <th>P>|z|</th>  <th>[0.025</th>    <th>0.975]</th>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>const</th> <td>   -0.1056</td> <td>    0.003</td> <td>  -42.118</td> <td> 0.000</td> <td>   -0.111</td> <td>   -0.101</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>x1</th>    <td>    0.0635</td> <td>    0.010</td> <td>    6.356</td> <td> 0.000</td> <td>    0.044</td> <td>    0.083</td>\n",
       "</tr>\n",
       "</table>"
      ],
      "text/plain": [
       "<class 'statsmodels.iolib.summary.Summary'>\n",
       "\"\"\"\n",
       "                           Logit Regression Results                           \n",
       "==============================================================================\n",
       "Dep. Variable:                outcome   No. Observations:               680519\n",
       "Model:                          Logit   Df Residuals:                   680517\n",
       "Method:                           MLE   Df Model:                            1\n",
       "Date:                Wed, 02 Nov 2022   Pseudo R-squ.:               4.288e-05\n",
       "Time:                        18:02:32   Log-Likelihood:            -4.7080e+05\n",
       "converged:                       True   LL-Null:                   -4.7082e+05\n",
       "Covariance Type:            nonrobust   LLR p-value:                 2.092e-10\n",
       "==============================================================================\n",
       "                 coef    std err          z      P>|z|      [0.025      0.975]\n",
       "------------------------------------------------------------------------------\n",
       "const         -0.1056      0.003    -42.118      0.000      -0.111      -0.101\n",
       "x1             0.0635      0.010      6.356      0.000       0.044       0.083\n",
       "==============================================================================\n",
       "\"\"\""
      ]
     },
     "execution_count": 204,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data=pd.DataFrame({'pre_post': results['pre_post'], 'initial_recovery':  results['initial_recovery_rates'],'outcome': results['outcomes'], 'limbic':results['limbic_used'] , 'number_referrals_pre':results['number_referral_initial'] , 'number_referrals_post':results['number_referral_final'] })\n",
    "data=data[data['pre_post']==1]\n",
    "x=np.transpose(np.array(data['limbic']))\n",
    "x=sm.add_constant(x)\n",
    "model = sm.Logit(data['outcome'], x).fit()\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "876e6148",
   "metadata": {},
   "source": [
    "## Create figure 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "id": "e1170f8b",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n"
     ]
    }
   ],
   "source": [
    "counts_recovery=[]\n",
    "for i, input_data in enumerate(inputs_services): \n",
    "    output=get_recovery_service(input_data)\n",
    "    print(output)\n",
    "    counts_recovery.append(output[4])\n",
    "counts_recovery=np.array(counts_recovery)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "id": "5b86cca2",
   "metadata": {},
   "outputs": [],
   "source": [
    "input_data_time={'start_date':'01/12/2020',\n",
    "    'final_date':'01/06/2022',\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "id": "54734c26",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n"
     ]
    }
   ],
   "source": [
    "counts_recovery_NHS=[]\n",
    "for i, input_data in enumerate(all_services): \n",
    "    if (input_data.lower().startswith('insight')==False) & (input_data.lower().startswith('trent')==False) & (input_data.lower().startswith('vita health')==False) & (input_data.lower().startswith('surrey iapt')==False)& (input_data.lower().startswith('all providers')==False): \n",
    "        output=get_recovery_service( {'service': input_data.lower(), 'start_date': input_data_time['start_date'], 'final_date': input_data_time['final_date']})\n",
    "    counts_recovery_NHS.append(output[4])\n",
    "counts_recovery_NHS=np.array(counts_recovery_NHS)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "id": "5992413d",
   "metadata": {},
   "outputs": [],
   "source": [
    "recovery_limbic_before=counts_recovery.sum(axis=0)[0]/(counts_recovery.sum(axis=0)[0]+counts_recovery.sum(axis=0)[1])\n",
    "recovery_limbic_after=counts_recovery.sum(axis=0)[2]/(counts_recovery.sum(axis=0)[2]+counts_recovery.sum(axis=0)[3])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "id": "2c9b73f9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(16.148428176321328,\n",
       " 5.8567152209312676e-05,\n",
       " 1,\n",
       " array([[ 7571.33461275,  8057.66538725],\n",
       "        [20700.66538725, 22030.33461275]]))"
      ]
     },
     "execution_count": 188,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "chi2_contingency([[counts_recovery.sum(axis=0)[0], counts_recovery.sum(axis=0)[1]], [counts_recovery.sum(axis=0)[2], counts_recovery.sum(axis=0)[3]]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "id": "cff8343a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(100.91358882018586,\n",
       " 9.608508201118352e-24,\n",
       " 1,\n",
       " array([[ 71427.08217268,  79827.91782732],\n",
       "        [352220.91782732, 393647.08217268]]))"
      ]
     },
     "execution_count": 189,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "chi2_contingency([[counts_recovery_NHS.sum(axis=0)[0], counts_recovery_NHS.sum(axis=0)[1]], [counts_recovery_NHS.sum(axis=0)[2], counts_recovery_NHS.sum(axis=0)[3]]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "id": "834448c7",
   "metadata": {},
   "outputs": [],
   "source": [
    "recovery_limbic_before=counts_recovery.sum(axis=0)[0]/(counts_recovery.sum(axis=0)[0]+counts_recovery.sum(axis=0)[1])\n",
    "recovery_limbic_after=counts_recovery.sum(axis=0)[2]/(counts_recovery.sum(axis=0)[2]+counts_recovery.sum(axis=0)[3])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "id": "7cb91f29",
   "metadata": {},
   "outputs": [],
   "source": [
    "recovery_NHS_before=counts_recovery_NHS.sum(axis=0)[0]/(counts_recovery_NHS.sum(axis=0)[0]+counts_recovery_NHS.sum(axis=0)[1])\n",
    "recovery_NHS_after=counts_recovery_NHS.sum(axis=0)[2]/(counts_recovery_NHS.sum(axis=0)[2]+counts_recovery_NHS.sum(axis=0)[3])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 208,
   "id": "e4688211",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "error_Limbic=[results_pre_post_limbic.std(axis=0)[0]/math.sqrt(len(results_pre_post_limbic)), results_pre_post_limbic.std(axis=0)[1]/math.sqrt(len(results_pre_post_limbic))]\n",
    "error_NHS=[results_pre_post_NHS.std(axis=0)[0]/math.sqrt(len(results_pre_post_NHS)), results_pre_post_NHS.std(axis=0)[1]/math.sqrt(len(results_pre_post_NHS))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 210,
   "id": "048afacc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([<matplotlib.axis.YTick at 0x25104d05970>,\n",
       "  <matplotlib.axis.YTick at 0x25104d05550>,\n",
       "  <matplotlib.axis.YTick at 0x25104d08f10>,\n",
       "  <matplotlib.axis.YTick at 0x25108dda070>,\n",
       "  <matplotlib.axis.YTick at 0x25108dda580>,\n",
       "  <matplotlib.axis.YTick at 0x25108ddaa90>],\n",
       " [Text(0, 0.4, '40'),\n",
       "  Text(0, 0.42, '42'),\n",
       "  Text(0, 0.44, '44'),\n",
       "  Text(0, 0.46, '46'),\n",
       "  Text(0, 0.48, '48'),\n",
       "  Text(0, 0.5, '50')])"
      ]
     },
     "execution_count": 210,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "fig, ax = plt.subplots(figsize=(10, 8))\n",
    "\n",
    "plt.bar([2,3], [recovery_NHS_before, recovery_NHS_after],yerr=error_NHS, color=(255/255,229/255,240/255) )\n",
    "plt.bar([0,1], [recovery_limbic_before, recovery_limbic_after],yerr=error_Limbic, color=(252/255,182/255,211/255))\n",
    "\n",
    "# plt.ylim([.4, .5])\n",
    "# plt.ylabel('recovery rates %', fontsize=20)\n",
    "\n",
    "plt.xticks([0,1,2,3], ['AI-tool pre', 'AI-tool post', 'NHS pre', 'NHS post'], fontsize=20)\n",
    "\n",
    "plt.ylim(.45, .51)\n",
    "plt.ylabel('recovery rates %', fontsize=20)\n",
    "plt.yticks([.4, .42, .44,.46, .48, .50], ['40','42', '44', '46', '48', '50'], fontsize=20)\n",
    "\n",
    "\n",
    "# image_format = 'svg' # e.g .png, .svg, etc.\n",
    "# image_name = folder_figures_save+'interaction.svg'\n",
    "\n",
    "# fig.savefig(image_name, format=image_format, dpi=1200)\n",
    "\n",
    "\n",
    "# image_format = 'png' # e.g .png, .svg, etc.\n",
    "# image_name = folder_figures_save+'Figure2.png'\n",
    "# fig.savefig(image_name, format=image_format, dpi=1200)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c5969972",
   "metadata": {},
   "source": [
    "## Define comparison group"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "id": "27803759",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "percentag_IAPT=50# 1=most similar IAPT service, 25=25% of NHS most similar IAPT service, 50=50% of NHS most similar IAPT services"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "38166404",
   "metadata": {},
   "source": [
    "## Sample similar IAPT services"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "id": "498265be",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n"
     ]
    }
   ],
   "source": [
    "\n",
    "text_trap = io.StringIO()\n",
    "sys.stdout = text_trap\n",
    "\n",
    "\n",
    "\n",
    "service_compare=[]\n",
    "distance_compare=[]\n",
    "direction_compare=[]\n",
    "output2=[]\n",
    "for i, input_data_limbic in enumerate(inputs_services): \n",
    "    vector_comparison=[]\n",
    "    service_id=[]\n",
    "    output=get_recovery_service(input_data_limbic)\n",
    "    output2=get_referral_service(input_data_limbic)\n",
    "    initial_recovery=output[0]\n",
    "    pre_numbers_treatment=output[4][0]+output[4][1]\n",
    "    post_numbers_treatment=output[4][2]+output[4][3]\n",
    "    try: \n",
    "        change=(post_numbers_treatment-pre_numbers_treatment)/(pre_numbers_treatment+pre_numbers_treatment)\n",
    "        change_referral=(output2[3]-output2[2])/(output2[2]+output2[3])\n",
    "    except: \n",
    "        change=0\n",
    "        change_referral=0\n",
    "\n",
    "    service_id.append(input_data_limbic['service'])\n",
    "    vector_comparison.append([initial_recovery, change, change_referral])\n",
    "\n",
    "\n",
    "    for i, input_data in enumerate(all_services): \n",
    "        if (input_data.lower().startswith('insight')==False) & (input_data.lower().startswith('trent')==False) & (input_data.lower().startswith('vita health')==False) & (input_data.lower().startswith('surrey iapt')==False)& (input_data.lower().startswith('all providers')==False): \n",
    "            output=get_recovery_service( {'service': input_data.lower(), 'start_date': input_data_limbic['start_date'], 'final_date': input_data_limbic['final_date']})\n",
    "            output2=get_referral_service( {'service': input_data.lower(), 'start_date': input_data_limbic['start_date'], 'final_date': input_data_limbic['final_date']})\n",
    "\n",
    "            initial_recovery=output[0]\n",
    "            if initial_recovery!=0: \n",
    "                pre_numbers_treatment=int(output[4][0]+output[4][1])\n",
    "                post_numbers_treatment=int(output[4][2]+output[4][3])\n",
    "                try: \n",
    "                    change=(post_numbers_treatment-pre_numbers_treatment)/(pre_numbers_treatment+pre_numbers_treatment)\n",
    "                    change_referral=(output2[3]-output2[2])/(output2[2]+output2[3])\n",
    "                except: \n",
    "                    change=0\n",
    "                service_id.append(input_data)\n",
    "                vector_comparison.append([initial_recovery, change, change_referral])\n",
    "\n",
    "        else: \n",
    "            print(input_data)\n",
    "    vector_comparison=np.array(vector_comparison)\n",
    "\n",
    "    normalised_vector_comparison=scipy.stats.zscore(vector_comparison, axis=0)\n",
    "    normalised_vector_comparison[:,0]=normalised_vector_comparison[:,0]\n",
    "\n",
    "    distance=[]\n",
    "    service=[]\n",
    "    input_matrix=[]\n",
    "    for i in range(normalised_vector_comparison.shape[0]): \n",
    "        # distance.append(cdist(vector_comparison[0, :2],vector_comparison[i, :2],'mahalanobis'))\n",
    "        distance.append(norm(normalised_vector_comparison[0, :2]-normalised_vector_comparison[i, :2]))\n",
    "\n",
    "        \n",
    "        # distance.append(norm(vector_comparison[0, 0]-vector_comparison[:, 0]))\n",
    "        service.append(service_id[i])\n",
    "        input_matrix.append(vector_comparison[i, :])\n",
    "    sorted_distance=pd.DataFrame({'service': service, 'distance':distance, 'input_matrix':input_matrix,\n",
    "    'direction':vector_comparison[0, 0]-vector_comparison[:, 0], \n",
    "    'initial_recovery':vector_comparison[:, 0], \n",
    "    'change_treatment':  vector_comparison[:, 1], \n",
    "    'change_referrals': vector_comparison[:, 2]\n",
    "     })\n",
    "    sorted_distance=sorted_distance.sort_values(by='distance')\n",
    "    if percentag_IAPT==50: \n",
    "        service_include=[sorted_distance['service'][sorted_distance['direction']>0].iloc[:46], sorted_distance['service'][sorted_distance['direction']<0].iloc[:46]]\n",
    "        distance_include=[sorted_distance['distance'][sorted_distance['direction']>0].iloc[:46], sorted_distance['distance'][sorted_distance['direction']<0].iloc[:46]]\n",
    "        direction_include=[sorted_distance['direction'][sorted_distance['direction']>0].iloc[:46], sorted_distance['direction'][sorted_distance['direction']<0].iloc[:46]]\n",
    "    elif percentag_IAPT==25:\n",
    "        service_include=[sorted_distance['service'][sorted_distance['direction']>0].iloc[:23], sorted_distance['service'][sorted_distance['direction']<0].iloc[:23]]\n",
    "        distance_include=[sorted_distance['distance'][sorted_distance['direction']>0].iloc[:23], sorted_distance['distance'][sorted_distance['direction']<0].iloc[:23]]\n",
    "        direction_include=[sorted_distance['direction'][sorted_distance['direction']>0].iloc[:23], sorted_distance['direction'][sorted_distance['direction']<0].iloc[:23]] \n",
    "    elif percentag_IAPT==1:\n",
    "        service_include=[sorted_distance['service'].iloc[1]]\n",
    "        distance_include=[sorted_distance['distance'].iloc[1]]\n",
    "        direction_include=[sorted_distance['direction'].iloc[1]]\n",
    "    if percentag_IAPT==1:\n",
    "        s_i=service_include\n",
    "        d_i=distance_include\n",
    "        dir_i=direction_include\n",
    "    else: \n",
    "        s_i=[]\n",
    "        for vec in service_include: \n",
    "            for serv in vec: \n",
    "                s_i.append(serv)\n",
    "        d_i=[]\n",
    "        for vec in distance_include: \n",
    "            for d in vec: \n",
    "                d_i.append(d)\n",
    "        dir_i=[]\n",
    "        for vec in direction_include: \n",
    "            for d in vec: \n",
    "                dir_i.append(d)\n",
    "    service_compare.append(s_i)\n",
    "    distance_compare.append(d_i)\n",
    "    direction_compare.append(dir_i)\n",
    "\n",
    "\n",
    "sys.stdout = sys.__stdout__\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "id": "55e57346",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n",
      "<ipython-input-69-c3d80b019509>:30: RuntimeWarning: invalid value encountered in true_divide\n",
      "  return  np.array(recovery).sum(), np.array(non_recovery).sum(), (np.array(recovery).sum()+np.array(non_recovery).sum())/len(dates_previous), np.array(recovery_timeline).sum(axis=0)/(np.array(recovery_timeline).sum(axis=0)+np.array(non_recovery_timeline).sum(axis=0))\n"
     ]
    }
   ],
   "source": [
    "import statsmodels.api as sm\n",
    "import scipy\n",
    "\n",
    "t_value_distribution=[]\n",
    "pre_NHS=[]\n",
    "pre_Limbic=[]\n",
    "NHS_referral=[]\n",
    "Limbic_referral=[]\n",
    "NHS_referral_change=[]\n",
    "Limbic_referral_change=[]\n",
    "\n",
    "for permutations in range(1):\n",
    "    results_pre_post_limbic=[]\n",
    "    results_pre_post_NHS=[]\n",
    "\n",
    "    recovery_pre=[]\n",
    "    outcomes=[]\n",
    "    pre_post=[]\n",
    "    service_id=[]\n",
    "    initial_recovery_rates=[]\n",
    "    limbic_used=[]\n",
    "    number_treatment_initial=[]\n",
    "    number_treatment_final=[]\n",
    "    waittime_limbic=[]\n",
    "    waittime_NHS=[]\n",
    "    waittime_diff=[]\n",
    "    service_code=[]\n",
    "    recovery_timeline_limbic_pre=[]\n",
    "    recovery_timeline_limbic_post=[]\n",
    "    serviceName_pre_post_NHS=[]\n",
    "\n",
    "    recovery_timeline_NHS_pre=[]\n",
    "    recovery_timeline_NHS_post=[]\n",
    "    used_comaprison=[]\n",
    "    output2=[]\n",
    "    referral_pre=[]\n",
    "    referral_post=[]\n",
    "    NHS_treatment_change=[]\n",
    "    Limbic_treatment_change=[]\n",
    "\n",
    "\n",
    "\n",
    "    for i, input_data in enumerate(inputs_services): \n",
    "        \n",
    "        output=get_recovery_service(input_data)\n",
    "        output2=get_referral_service(input_data)\n",
    "        results_pre_post_limbic.append([output[0], output[1]])\n",
    "        outcomes.append(prepare_output_vector(output[4]))\n",
    "        service_code.append([input_data]*int(output[4].sum()))\n",
    "        pre_post.append(prepare_pre_post(output[4]))\n",
    "        service_id.append(prepare_side_identifier(output[4],i))\n",
    "        initial_recovery_rates.append(prepare_side_identifier(output[4], output[0]))\n",
    "        limbic_used.append(prepare_side_identifier(output[4], 1))\n",
    "        number_treatment_initial.append(prepare_side_identifier(output[4], int(output[5])))\n",
    "        number_treatment_final.append(prepare_side_identifier(output[4], int(output[6])))\n",
    "        recovery_timeline_limbic_pre.append(output[7])\n",
    "        recovery_timeline_limbic_post.append(output[8])\n",
    "        referral_pre.append(prepare_side_identifier(output[4], int(output2[2])))\n",
    "        referral_post.append(prepare_side_identifier(output[4], int(output2[3])))\n",
    "    \n",
    "        # prob=np.array(1/np.array(distance_compare[i][:]))\n",
    "        # prob=prob/prob.sum()\n",
    "        # input_data2 =np.random.choice(service_compare[i][:],12,  p=prob)#,  p=prob\n",
    "        input_data2=service_compare[i]\n",
    "        for input_data in input_data2:\n",
    "            # while input_data in set(used_comaprison): \n",
    "            #     input_data =np.random.choice(service_compare[i][:],  p=prob) #, p=prob\n",
    "            used_comaprison.append(input_data)\n",
    "            output=get_recovery_service( {'service': input_data.lower(), 'start_date': input_data_limbic['start_date'], 'final_date': input_data_limbic['final_date']})\n",
    "            output2=get_referral_service( {'service': input_data.lower(), 'start_date': input_data_limbic['start_date'], 'final_date': input_data_limbic['final_date']})\n",
    "            service_code.append([input_data]*int(output[4].sum()))\n",
    "            if output[0]!=0: \n",
    "                results_pre_post_NHS.append([output[0], output[1]])\n",
    "            serviceName_pre_post_NHS.append(input_data)\n",
    "            outcomes.append(prepare_output_vector(output[4]))\n",
    "            pre_post.append(prepare_pre_post(output[4]))\n",
    "            service_id.append(prepare_side_identifier(output[4], i+4))\n",
    "            initial_recovery_rates.append(prepare_side_identifier(output[4], output[0]))\n",
    "            limbic_used.append(prepare_side_identifier(output[4], 0))\n",
    "            number_treatment_initial.append(prepare_side_identifier(output[4], int(output[5])))\n",
    "            number_treatment_final.append(prepare_side_identifier(output[4], int(output[6])))\n",
    "            recovery_timeline_NHS_pre.append(output[7])\n",
    "            recovery_timeline_NHS_post.append(output[8])\n",
    "            referral_pre.append(prepare_side_identifier(output[4], int(output2[2])))\n",
    "            referral_post.append(prepare_side_identifier(output[4], int(output2[3])))\n",
    "\n",
    "\n",
    "    outcomes=np.array(list(deepflatten(outcomes)))\n",
    "    pre_post=np.array(list(deepflatten(pre_post)))\n",
    "    service_id=np.array(list(deepflatten(service_id)))\n",
    "    initial_recovery_rates=np.array(list(deepflatten(initial_recovery_rates)))\n",
    "    limbic_used=np.array(list(deepflatten(limbic_used)))\n",
    "\n",
    "    service_code=np.array(list(deepflatten(service_code)))\n",
    "    number_treatment_initial=np.array(list(deepflatten(number_treatment_initial)))\n",
    "    number_treatment_final=np.array(list(deepflatten(number_treatment_final)))\n",
    "    referral_pre=np.array(list(deepflatten(referral_pre)))\n",
    "    referral_post=np.array(list(deepflatten(referral_post)))\n",
    "\n",
    "    data=pd.DataFrame({'pre_post': pre_post, 'initial_recovery': initial_recovery_rates, 'service': service_id, 'outcome': outcomes,\n",
    "     'limbic': limbic_used, 'number_referrals_pre': referral_pre, 'number_referrals_post': referral_post, \n",
    "     'number_treat_pre': number_treatment_initial, 'number_treat_post': number_treatment_final})\n",
    "\n",
    "    # x=np.concatenate((np.transpose(np.array([data['pre_post'], data['limbic'], data['pre_post']*data['limbic']])), categorical_service), axis=1)\n",
    "    x=np.transpose(np.array([data['pre_post'], data['limbic'], \n",
    "    data['pre_post']*data['limbic'], \n",
    "    data['initial_recovery'],\n",
    "    data['number_treat_post']-data['number_treat_pre']\n",
    "     ]))\n",
    "    x=sm.add_constant(x)\n",
    "    model = sm.Logit(data['outcome'], x).fit()\n",
    "    model.summary()\n",
    "    t_value_distribution.append(model.tvalues[3])\n",
    "    pre_NHS.append(data['outcome'][(data['pre_post']==0)&(data['limbic']==0)].mean())\n",
    "    pre_Limbic.append(data['outcome'][(data['pre_post']==0)&(data['limbic']==1)].mean())\n",
    "    NHS_referral.append(np.array(data['number_referrals_pre'][data['limbic']==0].unique()).mean())\n",
    "    Limbic_referral.append(np.array(data['number_referrals_pre'][data['limbic']==1].unique()).mean())\n",
    "    NHS_referral_change.append((np.array(data['number_referrals_post'][data['limbic']==0].unique()).mean()-np.array(data['number_referrals_pre'][data['limbic']==0].unique()).mean())/np.array(data['number_referrals_pre'][data['limbic']==0].unique()).mean())\n",
    "    Limbic_referral_change.append((np.array(data['number_referrals_post'][data['limbic']==1].unique()).mean()-np.array(data['number_referrals_pre'][data['limbic']==1].unique()).mean())/np.array(data['number_referrals_pre'][data['limbic']==1].unique()).mean())\n",
    "    NHS_treatment_change.append((np.array(data['number_treat_post'][data['limbic']==0].unique()).mean()-np.array(data['number_treat_pre'][data['limbic']==0].unique()).mean())/(np.array(data['number_treat_pre'][data['limbic']==1].unique()).mean()+np.array(data['number_treat_post'][data['limbic']==1].unique()).mean()))\n",
    "    Limbic_treatment_change.append((np.array(data['number_treat_post'][data['limbic']==1].unique()).mean()-np.array(data['number_treat_pre'][data['limbic']==1].unique()).mean())/(np.array(data['number_treat_pre'][data['limbic']==1].unique()).mean()+np.array(data['number_treat_post'][data['limbic']==1].unique()).mean()))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "id": "3b1b2320",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table class=\"simpletable\">\n",
       "<caption>Logit Regression Results</caption>\n",
       "<tr>\n",
       "  <th>Dep. Variable:</th>        <td>outcome</td>     <th>  No. Observations:  </th>   <td>1528424</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Model:</th>                 <td>Logit</td>      <th>  Df Residuals:      </th>   <td>1528418</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Method:</th>                 <td>MLE</td>       <th>  Df Model:          </th>   <td>     5</td>   \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Date:</th>            <td>Wed, 02 Nov 2022</td> <th>  Pseudo R-squ.:     </th>  <td>0.003498</td>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Time:</th>                <td>15:12:37</td>     <th>  Log-Likelihood:    </th> <td>-1.0532e+06</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>converged:</th>             <td>True</td>       <th>  LL-Null:           </th> <td>-1.0569e+06</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>Covariance Type:</th>     <td>nonrobust</td>    <th>  LLR p-value:       </th>   <td> 0.000</td>   \n",
       "</tr>\n",
       "</table>\n",
       "<table class=\"simpletable\">\n",
       "<tr>\n",
       "    <td></td>       <th>coef</th>     <th>std err</th>      <th>z</th>      <th>P>|z|</th>  <th>[0.025</th>    <th>0.975]</th>  \n",
       "</tr>\n",
       "<tr>\n",
       "  <th>const</th> <td>   -1.8157</td> <td>    0.021</td> <td>  -85.747</td> <td> 0.000</td> <td>   -1.857</td> <td>   -1.774</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>x1</th>    <td>   -0.0572</td> <td>    0.004</td> <td>  -15.957</td> <td> 0.000</td> <td>   -0.064</td> <td>   -0.050</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>x2</th>    <td>   -0.0036</td> <td>    0.016</td> <td>   -0.223</td> <td> 0.823</td> <td>   -0.036</td> <td>    0.028</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>x3</th>    <td>    0.1602</td> <td>    0.019</td> <td>    8.399</td> <td> 0.000</td> <td>    0.123</td> <td>    0.198</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>x4</th>    <td>   -0.0006</td> <td>    0.003</td> <td>   -0.205</td> <td> 0.837</td> <td>   -0.006</td> <td>    0.005</td>\n",
       "</tr>\n",
       "<tr>\n",
       "  <th>x5</th>    <td>    3.6156</td> <td>    0.044</td> <td>   83.026</td> <td> 0.000</td> <td>    3.530</td> <td>    3.701</td>\n",
       "</tr>\n",
       "</table>"
      ],
      "text/plain": [
       "<class 'statsmodels.iolib.summary.Summary'>\n",
       "\"\"\"\n",
       "                           Logit Regression Results                           \n",
       "==============================================================================\n",
       "Dep. Variable:                outcome   No. Observations:              1528424\n",
       "Model:                          Logit   Df Residuals:                  1528418\n",
       "Method:                           MLE   Df Model:                            5\n",
       "Date:                Wed, 02 Nov 2022   Pseudo R-squ.:                0.003498\n",
       "Time:                        15:12:37   Log-Likelihood:            -1.0532e+06\n",
       "converged:                       True   LL-Null:                   -1.0569e+06\n",
       "Covariance Type:            nonrobust   LLR p-value:                     0.000\n",
       "==============================================================================\n",
       "                 coef    std err          z      P>|z|      [0.025      0.975]\n",
       "------------------------------------------------------------------------------\n",
       "const         -1.8157      0.021    -85.747      0.000      -1.857      -1.774\n",
       "x1            -0.0572      0.004    -15.957      0.000      -0.064      -0.050\n",
       "x2            -0.0036      0.016     -0.223      0.823      -0.036       0.028\n",
       "x3             0.1602      0.019      8.399      0.000       0.123       0.198\n",
       "x4            -0.0006      0.003     -0.205      0.837      -0.006       0.005\n",
       "x5             3.6156      0.044     83.026      0.000       3.530       3.701\n",
       "==============================================================================\n",
       "\"\"\""
      ]
     },
     "execution_count": 182,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x=np.transpose(np.array([data['pre_post'], data['limbic'], \n",
    "data['pre_post']*data['limbic'], \n",
    "(data['number_treat_post']-data['number_treat_pre'])/data['number_treat_pre'], \n",
    "data['initial_recovery']\n",
    " ]))\n",
    "x=sm.add_constant(x)\n",
    "model = sm.Logit(data['outcome'], x).fit()\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "id": "8b9bfdbf",
   "metadata": {},
   "outputs": [],
   "source": [
    "recovery_NHS_before=np.array(results_pre_post_NHS).mean(axis=0)[0]\n",
    "recovery_NHS_after=np.array(results_pre_post_NHS).mean(axis=0)[1]\n",
    "error_NHS=np.array(results_pre_post_NHS).std(axis=0)/math.sqrt(len(results_pre_post_NHS))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "id": "8e99c8ce",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "fig, ax = plt.subplots(figsize=(10, 8))\n",
    "\n",
    "\n",
    "plt.bar([2,3], [recovery_NHS_before, recovery_NHS_after],yerr=error_NHS,color=(255/255,229/255,240/255) )\n",
    "plt.bar([0,1], [recovery_limbic_before, recovery_limbic_after],yerr=error_Limbic, color=(252/255,182/255,211/255))\n",
    "\n",
    "\n",
    "plt.xticks([0,1,2,3], ['AI-tool pre', 'AI-tool post', 'NHS pre', 'NHS post'], fontsize=20)\n",
    "\n",
    "plt.ylim(.45, .51)\n",
    "plt.ylabel('recovery rates %', fontsize=20)\n",
    "plt.yticks([.4, .42, .44,.46, .48, .50], ['40','42', '44', '46', '48', '50'], fontsize=20)\n",
    "\n",
    "\n",
    "image_format = 'svg' # e.g .png, .svg, etc.\n",
    "image_name = folder_figures_save+'NHS_50.svg'\n",
    "\n",
    "fig.savefig(image_name, format=image_format, dpi=1200)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3f0e1a81",
   "metadata": {},
   "source": [
    "## Conduct economic analysis\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "id": "6f23b47c",
   "metadata": {},
   "outputs": [],
   "source": [
    "improvements_limbic=(recovery_NHS_before-recovery_NHS_after)+(recovery_limbic_after-recovery_limbic_before)\n",
    "number_improved_referrals=round((counts_recovery[:, 2:].sum())*(improvements_limbic))#\n",
    "number_referred_patients=27029\n",
    "price_point=[5, 7.5, 10]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "id": "3298e711",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>price per referral</th>\n",
       "      <th>price overall</th>\n",
       "      <th>price per additional recovery</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.0</td>\n",
       "      <td>135145.0</td>\n",
       "      <td>103.638804</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>7.5</td>\n",
       "      <td>202717.5</td>\n",
       "      <td>155.458206</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10.0</td>\n",
       "      <td>270290.0</td>\n",
       "      <td>207.277607</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   price per referral  price overall  price per additional recovery\n",
       "0                 5.0       135145.0                     103.638804\n",
       "1                 7.5       202717.5                     155.458206\n",
       "2                10.0       270290.0                     207.277607"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "price_overall=[]\n",
    "price_per_recovery=[]\n",
    "for p in price_point: \n",
    "    price_overall.append(number_referred_patients*p)\n",
    "    price_per_recovery.append((number_referred_patients*p)/number_improved_referrals)\n",
    "cost_Limbic=pd.DataFrame({'price per referral': price_point, 'price overall': price_overall, 'price per additional recovery': price_per_recovery})\n",
    "cost_Limbic"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fe1faf05",
   "metadata": {},
   "source": [
    "### Comparison to alternatives to improve recoveries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "id": "311b0338",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1021"
      ]
     },
     "execution_count": 99,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cost_single_IAPT_session_low_intensity=102.38 # Griffith & Steen \n",
    "cost_single_IAPT_session_high_intensity=173.88 #  Griffith & Steen \n",
    "odds_ratio_session=math.exp(0.186) #Gyian et al show that an additional session has the a beta = .186 on recovery rates\n",
    "percentage_improved_session=(odds_ratio_session*recovery_limbic_before)/(1+(odds_ratio_session*recovery_limbic_before)-recovery_limbic_before)\n",
    "number_improved_session=round((counts_recovery.sum())*((percentage_improved_session)-(recovery_limbic_before)))\n",
    "total_cost_session=round(number_referred_patients*cost_single_IAPT_session_low_intensity)\n",
    "\n",
    "cost_per_recovery_session=round((number_referred_patients*cost_single_IAPT_session_low_intensity)/number_improved_session)\n",
    "cost_per_recovery_session"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "72b6b518",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.9.7 64-bit",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.7"
  },
  "vscode": {
   "interpreter": {
    "hash": "ad8259fb0131ae414e0fed75b17e89163033ec83dd8a20ff234b9709c1123196"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}