Parkinsons-Predictor.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "6d280140",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Defaulting to user installation because normal site-packages is not writeable\n",
      "Collecting xgboost\n",
      "  Downloading xgboost-1.7.5-py3-none-macosx_12_0_arm64.whl (1.6 MB)\n",
      "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m1.6/1.6 MB\u001b[0m \u001b[31m8.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0ma \u001b[36m0:00:01\u001b[0m\n",
      "\u001b[?25hRequirement already satisfied: scipy in /Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/site-packages (from xgboost) (1.9.1)\n",
      "Requirement already satisfied: numpy in /Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/site-packages (from xgboost) (1.23.2)\n",
      "Installing collected packages: xgboost\n",
      "Successfully installed xgboost-1.7.5\n",
      "\n",
      "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m A new release of pip is available: \u001b[0m\u001b[31;49m23.0.1\u001b[0m\u001b[39;49m -> \u001b[0m\u001b[32;49m23.1.2\u001b[0m\n",
      "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m To update, run: \u001b[0m\u001b[32;49mpip3 install --upgrade pip\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "#!pip3 install xgboost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "dabc0e4f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Defaulting to user installation because normal site-packages is not writeable\n",
      "Collecting kaggle\n",
      "  Downloading kaggle-1.5.13.tar.gz (63 kB)\n",
      "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m63.3/63.3 kB\u001b[0m \u001b[31m926.4 kB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m \u001b[36m0:00:01\u001b[0m\n",
      "\u001b[?25h  Preparing metadata (setup.py) ... \u001b[?25ldone\n",
      "\u001b[?25hRequirement already satisfied: six>=1.10 in /Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/site-packages (from kaggle) (1.16.0)\n",
      "Requirement already satisfied: certifi in /Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/site-packages (from kaggle) (2022.6.15)\n",
      "Requirement already satisfied: python-dateutil in /Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/site-packages (from kaggle) (2.8.2)\n",
      "Requirement already satisfied: requests in /Users/datascience/Library/Python/3.10/lib/python/site-packages (from kaggle) (2.29.0)\n",
      "Requirement already satisfied: tqdm in /Users/datascience/Library/Python/3.10/lib/python/site-packages (from kaggle) (4.65.0)\n",
      "Collecting python-slugify\n",
      "  Downloading python_slugify-8.0.1-py2.py3-none-any.whl (9.7 kB)\n",
      "Requirement already satisfied: urllib3 in /Users/datascience/Library/Python/3.10/lib/python/site-packages (from kaggle) (1.26.15)\n",
      "Collecting text-unidecode>=1.3\n",
      "  Downloading text_unidecode-1.3-py2.py3-none-any.whl (78 kB)\n",
      "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m78.2/78.2 kB\u001b[0m \u001b[31m2.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
      "\u001b[?25hRequirement already satisfied: charset-normalizer<4,>=2 in /Users/datascience/Library/Python/3.10/lib/python/site-packages (from requests->kaggle) (3.1.0)\n",
      "Requirement already satisfied: idna<4,>=2.5 in /Users/datascience/Library/Python/3.10/lib/python/site-packages (from requests->kaggle) (3.4)\n",
      "Installing collected packages: text-unidecode, python-slugify, kaggle\n",
      "\u001b[33m  WARNING: The script slugify is installed in '/Users/datascience/Library/Python/3.10/bin' which is not on PATH.\n",
      "  Consider adding this directory to PATH or, if you prefer to suppress this warning, use --no-warn-script-location.\u001b[0m\u001b[33m\n",
      "\u001b[0m\u001b[33m  DEPRECATION: kaggle is being installed using the legacy 'setup.py install' method, because it does not have a 'pyproject.toml' and the 'wheel' package is not installed. pip 23.1 will enforce this behaviour change. A possible replacement is to enable the '--use-pep517' option. Discussion can be found at https://github.com/pypa/pip/issues/8559\u001b[0m\u001b[33m\n",
      "\u001b[0m  Running setup.py install for kaggle ... \u001b[?25ldone\n",
      "\u001b[?25hSuccessfully installed kaggle-1.5.13 python-slugify-8.0.1 text-unidecode-1.3\n",
      "\n",
      "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m A new release of pip is available: \u001b[0m\u001b[31;49m23.0.1\u001b[0m\u001b[39;49m -> \u001b[0m\u001b[32;49m23.1.2\u001b[0m\n",
      "\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m To update, run: \u001b[0m\u001b[32;49mpip3 install --upgrade pip\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "#!pip3 install kaggle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f831761a",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "id": "bb5645b1",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "import numpy as np #Linear Algebra \n",
    "import pandas as pd #Data Analysis \n",
    "import matplotlib.pyplot as plt\n",
    "from matplotlib.lines import Line2D\n",
    "import seaborn as sns\n",
    "\n",
    "\n",
    "from sklearn import svm #Classification / Regression Identification tasks\n",
    "from sklearn.metrics import accuracy_score #Calculates accuracy score \n",
    "from sklearn.preprocessing import StandardScaler#Estimates Empirical Mean & standard deviation\n",
    "from sklearn.model_selection import train_test_split #Takes separate datasets for training & testing purposes respectively\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from xgboost import XGBRegressor\n",
    "\n",
    "import os\n",
    "for dirname, _, filenames in os.walk('/kaggle/input'):\n",
    "    for filename in filenames:\n",
    "        print(os.path.join(dirname, filename))\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b6a9f365",
   "metadata": {},
   "source": [
    "# About Files\n",
    "\n",
    " train_peptides.csv Mass spectrometry data at the peptide level. Peptides are the component subunits of proteins.\n",
    "\n",
    "1. visit_id - ID code for the visit.\n",
    "1. visit_month - The month of the visit, relative to the first visit by the patient.\n",
    "1. patient_id - An ID code for the patient.\n",
    "1. UniProt - The UniProt ID code for the associated protein. There are often several peptides per protein.\n",
    "1. Peptide - The sequence of amino acids included in the peptide. See this table for the relevant codes. Some rare annotations may not be included in the table. The test set may include peptides not found in the train set.\n",
    "1. PeptideAbundance - The frequency of the amino acid in the sample.\n",
    "\n",
    "\n",
    "train_proteins.csv Protein expression frequencies aggregated from the peptide level data.\n",
    "\n",
    "1. visit_id - ID code for the visit.\n",
    "1. visit_month - The month of the visit, relative to the first visit by the patient.\n",
    "1. patient_id - An ID code for the patient.\n",
    "1. UniProt - The UniProt ID code for the associated protein. There are often several peptides per protein. The test set may include proteins not found in the train set.\n",
    "1. NPX - Normalized protein expression. The frequency of the protein's occurrence in the sample. May not have a 1:1 relationship with the component peptides as some proteins contain repeated copies of a given peptide.\n",
    "\n",
    "train_clinical_data.csv\n",
    "\n",
    "1. visit_id - ID code for the visit.\n",
    "1. visit_month - The month of the visit, relative to the first visit by the patient.\n",
    "1. patient_id - An ID code for the patient. \n",
    "1. updrs_[1-4] - The patient's score for part N of the Unified Parkinson's Disease Rating Scale. Higher numbers indicate more severe symptoms. Each sub-section covers a distinct category of symptoms, such as mood and behavior for Part 1 and motor functions for Part 3.\n",
    "1. upd23b_clinical_state_on_medication - Whether or not the patient was taking medication such as Levodopa during the UPDRS assessment. Expected to mainly affect the scores for Part 3 (motor function). These medications wear off fairly quickly (on the order of one day) so it's common for patients to take the motor function exam twice in a single month, both with and without medication.\n",
    "1. supplemental_clinical_data.csv Clinical records without any associated CSF samples. This data is intended to provide additional context about the typical progression of Parkinsons. Uses the same columns as train_clinical_data.csv.\n",
    "\n",
    "example_test_files/ Data intended to illustrate how the API functions. Includes the same columns delivered by the API (ie no updrs columns).\n",
    "\n",
    "amp_pd_peptide/ Files that enable the API. Expect the API to deliver all of the data (less than 1,000 additional patients) in under five minutes and to reserve less than 0.5 GB of memory. A brief demonstration of what the API delivers is available here.\n",
    "\n",
    "public_timeseries_testing_util.py An optional file intended to make it easier to run custom offline API tests. See the script's docstring for details.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f56ff89e",
   "metadata": {},
   "source": [
    "# Reading Files"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "099fefe7",
   "metadata": {},
   "outputs": [],
   "source": [
    "proteins_data = pd.read_csv(\"train_proteins.csv\")\n",
    "peptides_data = pd.read_csv(\"train_peptides.csv\")\n",
    "clinical_data = pd.read_csv(\"train_clinical_data.csv\")\n",
    "clinical_sup_data = pd.read_csv(\"supplemental_clinical_data.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "e6c6f423",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(2615, 8)\n",
      "(981834, 6)\n",
      "(232741, 5)\n"
     ]
    }
   ],
   "source": [
    "print(df_train_clin.shape)\n",
    "print(df_train_pept.shape)\n",
    "print(df_train_prot.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "4b6ceab3",
   "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>visit_id</th>\n",
       "      <th>patient_id</th>\n",
       "      <th>visit_month</th>\n",
       "      <th>updrs_1</th>\n",
       "      <th>updrs_2</th>\n",
       "      <th>updrs_3</th>\n",
       "      <th>updrs_4</th>\n",
       "      <th>upd23b_clinical_state_on_medication</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>55_0</td>\n",
       "      <td>55</td>\n",
       "      <td>0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>55_3</td>\n",
       "      <td>55</td>\n",
       "      <td>3</td>\n",
       "      <td>10.0</td>\n",
       "      <td>7.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>55_6</td>\n",
       "      <td>55</td>\n",
       "      <td>6</td>\n",
       "      <td>8.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>34.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>55_9</td>\n",
       "      <td>55</td>\n",
       "      <td>9</td>\n",
       "      <td>8.0</td>\n",
       "      <td>9.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>On</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>55_12</td>\n",
       "      <td>55</td>\n",
       "      <td>12</td>\n",
       "      <td>10.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>41.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>On</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  visit_id  patient_id  visit_month  updrs_1  updrs_2  updrs_3  updrs_4  \\\n",
       "0     55_0          55            0     10.0      6.0     15.0      NaN   \n",
       "1     55_3          55            3     10.0      7.0     25.0      NaN   \n",
       "2     55_6          55            6      8.0     10.0     34.0      NaN   \n",
       "3     55_9          55            9      8.0      9.0     30.0      0.0   \n",
       "4    55_12          55           12     10.0     10.0     41.0      0.0   \n",
       "\n",
       "  upd23b_clinical_state_on_medication  \n",
       "0                                 NaN  \n",
       "1                                 NaN  \n",
       "2                                 NaN  \n",
       "3                                  On  \n",
       "4                                  On  "
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#We will start with the Clinical Data\n",
    "clinical_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "c18d74ee",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "proteins shape:\t\t (232741, 5)\n",
      "peptides shape:\t\t (981834, 6)\n",
      "clinical shape:\t\t (2615, 8)\n",
      "sup clinical shape:\t (2223, 8)\n"
     ]
    }
   ],
   "source": [
    "#dataset shape\n",
    "print('proteins shape:\\t\\t', proteins_data.shape)\n",
    "print('peptides shape:\\t\\t', peptides_data.shape)\n",
    "print('clinical shape:\\t\\t', clinical_data.shape)\n",
    "print('sup clinical shape:\\t', clinical_sup_data.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "6aaba2f8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Unique patient_ids in clinical_data : 248\n",
      "Unique visit_ids in testing clinical_data : 2615\n",
      "Unique visit_months in clinical_data: 17\n"
     ]
    }
   ],
   "source": [
    "#Finding Num of Unique Entries In Each Data Set\n",
    "print(\"Unique patient_ids in clinical_data :\", len(clinical_data[\"patient_id\"].unique()))\n",
    "print(\"Unique visit_ids in testing clinical_data :\", len(clinical_data[\"visit_id\"].unique()))\n",
    "print(\"Unique visit_months in clinical_data:\", len(clinical_data[\"visit_month\"].unique()))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "ef2938e7",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Unique patient_ids in clinical_sup : 771\n",
      "Unique visit_ids in clinical_sup : 2223\n",
      "Unique visit_months in clinical_sup : 17\n"
     ]
    }
   ],
   "source": [
    "print(\"Unique patient_ids in clinical_sup :\", len(clinical_sup_data[\"patient_id\"].unique()))\n",
    "print(\"Unique visit_ids in clinical_sup :\", len(clinical_sup_data[\"visit_id\"].unique()))\n",
    "print(\"Unique visit_months in clinical_sup :\", len(clinical_data[\"visit_month\"].unique()))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "87737547",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Unique patient_ids in proteins_data : 248\n",
      "Unique visit_ids in proteins_data : 1113\n",
      "Unique visit_months in proteins_data : 15\n",
      "Unique UniProt in proteins_data : 227\n",
      "Unique NPX in proteins_data : 218795\n"
     ]
    }
   ],
   "source": [
    "print(\"Unique patient_ids in proteins_data :\", len(proteins_data[\"patient_id\"].unique()))\n",
    "print(\"Unique visit_ids in proteins_data :\", len(proteins_data[\"visit_id\"].unique()))\n",
    "print(\"Unique visit_months in proteins_data :\", len(proteins_data[\"visit_month\"].unique()))\n",
    "print(\"Unique UniProt in proteins_data :\", len(proteins_data[\"UniProt\"].unique()))\n",
    "print(\"Unique NPX in proteins_data :\", len(proteins_data[\"NPX\"].unique()))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5bb2e496",
   "metadata": {},
   "source": [
    "# Data Merging\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "e0b61efd",
   "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>visit_id</th>\n",
       "      <th>visit_month</th>\n",
       "      <th>patient_id</th>\n",
       "      <th>O00391</th>\n",
       "      <th>O00533</th>\n",
       "      <th>O00584</th>\n",
       "      <th>O14498</th>\n",
       "      <th>O14773</th>\n",
       "      <th>O14791</th>\n",
       "      <th>O15240</th>\n",
       "      <th>...</th>\n",
       "      <th>Q9HDC9</th>\n",
       "      <th>Q9NQ79</th>\n",
       "      <th>Q9NYU2</th>\n",
       "      <th>Q9UBR2</th>\n",
       "      <th>Q9UBX5</th>\n",
       "      <th>Q9UHG2</th>\n",
       "      <th>Q9UKV8</th>\n",
       "      <th>Q9UNU6</th>\n",
       "      <th>Q9Y646</th>\n",
       "      <th>Q9Y6R7</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10053_0</td>\n",
       "      <td>0</td>\n",
       "      <td>10053</td>\n",
       "      <td>9104.27</td>\n",
       "      <td>402321.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>7150.57</td>\n",
       "      <td>2497.84</td>\n",
       "      <td>83002.9</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>9469.45</td>\n",
       "      <td>94237.6</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23016.0</td>\n",
       "      <td>177983.0</td>\n",
       "      <td>65900.0</td>\n",
       "      <td>15382.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>19017.40</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10053_12</td>\n",
       "      <td>12</td>\n",
       "      <td>10053</td>\n",
       "      <td>10464.20</td>\n",
       "      <td>435586.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>197117.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>14408.40</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>28537.0</td>\n",
       "      <td>171733.0</td>\n",
       "      <td>65668.1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>9295.65</td>\n",
       "      <td>25697.80</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10053_18</td>\n",
       "      <td>18</td>\n",
       "      <td>10053</td>\n",
       "      <td>13235.70</td>\n",
       "      <td>507386.0</td>\n",
       "      <td>7126.96</td>\n",
       "      <td>24525.7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2372.71</td>\n",
       "      <td>126506.0</td>\n",
       "      <td>...</td>\n",
       "      <td>317477.0</td>\n",
       "      <td>38667.20</td>\n",
       "      <td>111107.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>37932.6</td>\n",
       "      <td>245188.0</td>\n",
       "      <td>59986.1</td>\n",
       "      <td>10813.3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>29102.70</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10138_12</td>\n",
       "      <td>12</td>\n",
       "      <td>10138</td>\n",
       "      <td>12600.20</td>\n",
       "      <td>494581.0</td>\n",
       "      <td>9165.06</td>\n",
       "      <td>27193.5</td>\n",
       "      <td>22506.10</td>\n",
       "      <td>6015.90</td>\n",
       "      <td>156313.0</td>\n",
       "      <td>...</td>\n",
       "      <td>557904.0</td>\n",
       "      <td>44556.90</td>\n",
       "      <td>155619.0</td>\n",
       "      <td>14647.9</td>\n",
       "      <td>36927.7</td>\n",
       "      <td>229232.0</td>\n",
       "      <td>106564.0</td>\n",
       "      <td>26077.7</td>\n",
       "      <td>21441.80</td>\n",
       "      <td>7642.42</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10138_24</td>\n",
       "      <td>24</td>\n",
       "      <td>10138</td>\n",
       "      <td>12003.20</td>\n",
       "      <td>522138.0</td>\n",
       "      <td>4498.51</td>\n",
       "      <td>17189.8</td>\n",
       "      <td>29112.40</td>\n",
       "      <td>2665.15</td>\n",
       "      <td>151169.0</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>47836.70</td>\n",
       "      <td>177619.0</td>\n",
       "      <td>17061.1</td>\n",
       "      <td>25510.4</td>\n",
       "      <td>176722.0</td>\n",
       "      <td>59471.4</td>\n",
       "      <td>12639.2</td>\n",
       "      <td>15091.40</td>\n",
       "      <td>6168.55</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 230 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   visit_id  visit_month  patient_id    O00391    O00533   O00584   O14498  \\\n",
       "0   10053_0            0       10053   9104.27  402321.0      NaN      NaN   \n",
       "1  10053_12           12       10053  10464.20  435586.0      NaN      NaN   \n",
       "2  10053_18           18       10053  13235.70  507386.0  7126.96  24525.7   \n",
       "3  10138_12           12       10138  12600.20  494581.0  9165.06  27193.5   \n",
       "4  10138_24           24       10138  12003.20  522138.0  4498.51  17189.8   \n",
       "\n",
       "     O14773   O14791    O15240  ...    Q9HDC9    Q9NQ79    Q9NYU2   Q9UBR2  \\\n",
       "0   7150.57  2497.84   83002.9  ...       NaN   9469.45   94237.6      NaN   \n",
       "1       NaN      NaN  197117.0  ...       NaN  14408.40       NaN      NaN   \n",
       "2       NaN  2372.71  126506.0  ...  317477.0  38667.20  111107.0      NaN   \n",
       "3  22506.10  6015.90  156313.0  ...  557904.0  44556.90  155619.0  14647.9   \n",
       "4  29112.40  2665.15  151169.0  ...       NaN  47836.70  177619.0  17061.1   \n",
       "\n",
       "    Q9UBX5    Q9UHG2    Q9UKV8   Q9UNU6    Q9Y646    Q9Y6R7  \n",
       "0  23016.0  177983.0   65900.0  15382.0       NaN  19017.40  \n",
       "1  28537.0  171733.0   65668.1      NaN   9295.65  25697.80  \n",
       "2  37932.6  245188.0   59986.1  10813.3       NaN  29102.70  \n",
       "3  36927.7  229232.0  106564.0  26077.7  21441.80   7642.42  \n",
       "4  25510.4  176722.0   59471.4  12639.2  15091.40   6168.55  \n",
       "\n",
       "[5 rows x 230 columns]"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Dropping redundant information in the Protein set.\n",
    "proteins_merge = proteins_data.pivot(index=['visit_id', 'visit_month', 'patient_id'], \n",
    "          columns='UniProt',\n",
    "          values='NPX').reset_index().rename_axis(None, axis=1)\n",
    "\n",
    "proteins_merge.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "49ea1558",
   "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>visit_id</th>\n",
       "      <th>O15240</th>\n",
       "      <th>P01009</th>\n",
       "      <th>P01011</th>\n",
       "      <th>P01023</th>\n",
       "      <th>P01024</th>\n",
       "      <th>P01042</th>\n",
       "      <th>P01834</th>\n",
       "      <th>P01876</th>\n",
       "      <th>P02647</th>\n",
       "      <th>...</th>\n",
       "      <th>P02790</th>\n",
       "      <th>P05090</th>\n",
       "      <th>P06396</th>\n",
       "      <th>P07602</th>\n",
       "      <th>P10909</th>\n",
       "      <th>P23142</th>\n",
       "      <th>P41222</th>\n",
       "      <th>Q12805</th>\n",
       "      <th>Q92520</th>\n",
       "      <th>Q9UHG2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10053_0</td>\n",
       "      <td>83002.9</td>\n",
       "      <td>12825300.0</td>\n",
       "      <td>1084770.0</td>\n",
       "      <td>1005230.0</td>\n",
       "      <td>2669740.0</td>\n",
       "      <td>541909.0</td>\n",
       "      <td>2440800.0</td>\n",
       "      <td>1874270.0</td>\n",
       "      <td>4032650.0</td>\n",
       "      <td>...</td>\n",
       "      <td>3858400.0</td>\n",
       "      <td>4087060.0</td>\n",
       "      <td>1352720.0</td>\n",
       "      <td>731854.0</td>\n",
       "      <td>4749780.0</td>\n",
       "      <td>822784.0</td>\n",
       "      <td>42511700.0</td>\n",
       "      <td>1051980.0</td>\n",
       "      <td>322496.0</td>\n",
       "      <td>177983.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10053_12</td>\n",
       "      <td>197117.0</td>\n",
       "      <td>11871200.0</td>\n",
       "      <td>1135570.0</td>\n",
       "      <td>1230990.0</td>\n",
       "      <td>3360790.0</td>\n",
       "      <td>584371.0</td>\n",
       "      <td>2030500.0</td>\n",
       "      <td>1800070.0</td>\n",
       "      <td>5111760.0</td>\n",
       "      <td>...</td>\n",
       "      <td>3708010.0</td>\n",
       "      <td>3491420.0</td>\n",
       "      <td>1390950.0</td>\n",
       "      <td>618772.0</td>\n",
       "      <td>4223100.0</td>\n",
       "      <td>661058.0</td>\n",
       "      <td>34100500.0</td>\n",
       "      <td>738247.0</td>\n",
       "      <td>496976.0</td>\n",
       "      <td>171733.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10053_18</td>\n",
       "      <td>126506.0</td>\n",
       "      <td>14319900.0</td>\n",
       "      <td>1444430.0</td>\n",
       "      <td>1333940.0</td>\n",
       "      <td>3728860.0</td>\n",
       "      <td>971184.0</td>\n",
       "      <td>2106740.0</td>\n",
       "      <td>1508300.0</td>\n",
       "      <td>5713180.0</td>\n",
       "      <td>...</td>\n",
       "      <td>4773570.0</td>\n",
       "      <td>5096710.0</td>\n",
       "      <td>2232400.0</td>\n",
       "      <td>879126.0</td>\n",
       "      <td>5728010.0</td>\n",
       "      <td>971421.0</td>\n",
       "      <td>38324600.0</td>\n",
       "      <td>843926.0</td>\n",
       "      <td>674551.0</td>\n",
       "      <td>245188.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10138_12</td>\n",
       "      <td>156313.0</td>\n",
       "      <td>11024900.0</td>\n",
       "      <td>1725440.0</td>\n",
       "      <td>1302590.0</td>\n",
       "      <td>4276570.0</td>\n",
       "      <td>1627070.0</td>\n",
       "      <td>4071840.0</td>\n",
       "      <td>5456100.0</td>\n",
       "      <td>22825800.0</td>\n",
       "      <td>...</td>\n",
       "      <td>6629400.0</td>\n",
       "      <td>5233080.0</td>\n",
       "      <td>2535650.0</td>\n",
       "      <td>1204390.0</td>\n",
       "      <td>7528540.0</td>\n",
       "      <td>1246230.0</td>\n",
       "      <td>34148900.0</td>\n",
       "      <td>816347.0</td>\n",
       "      <td>601869.0</td>\n",
       "      <td>229232.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10138_24</td>\n",
       "      <td>151169.0</td>\n",
       "      <td>14145200.0</td>\n",
       "      <td>1544440.0</td>\n",
       "      <td>1221060.0</td>\n",
       "      <td>4570310.0</td>\n",
       "      <td>1301680.0</td>\n",
       "      <td>2421520.0</td>\n",
       "      <td>3032880.0</td>\n",
       "      <td>11136600.0</td>\n",
       "      <td>...</td>\n",
       "      <td>2628690.0</td>\n",
       "      <td>3852520.0</td>\n",
       "      <td>2443370.0</td>\n",
       "      <td>459165.0</td>\n",
       "      <td>8704760.0</td>\n",
       "      <td>1274950.0</td>\n",
       "      <td>45982800.0</td>\n",
       "      <td>598882.0</td>\n",
       "      <td>569973.0</td>\n",
       "      <td>176722.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 28 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   visit_id    O15240      P01009     P01011     P01023     P01024     P01042  \\\n",
       "0   10053_0   83002.9  12825300.0  1084770.0  1005230.0  2669740.0   541909.0   \n",
       "1  10053_12  197117.0  11871200.0  1135570.0  1230990.0  3360790.0   584371.0   \n",
       "2  10053_18  126506.0  14319900.0  1444430.0  1333940.0  3728860.0   971184.0   \n",
       "3  10138_12  156313.0  11024900.0  1725440.0  1302590.0  4276570.0  1627070.0   \n",
       "4  10138_24  151169.0  14145200.0  1544440.0  1221060.0  4570310.0  1301680.0   \n",
       "\n",
       "      P01834     P01876      P02647  ...     P02790     P05090     P06396  \\\n",
       "0  2440800.0  1874270.0   4032650.0  ...  3858400.0  4087060.0  1352720.0   \n",
       "1  2030500.0  1800070.0   5111760.0  ...  3708010.0  3491420.0  1390950.0   \n",
       "2  2106740.0  1508300.0   5713180.0  ...  4773570.0  5096710.0  2232400.0   \n",
       "3  4071840.0  5456100.0  22825800.0  ...  6629400.0  5233080.0  2535650.0   \n",
       "4  2421520.0  3032880.0  11136600.0  ...  2628690.0  3852520.0  2443370.0   \n",
       "\n",
       "      P07602     P10909     P23142      P41222     Q12805    Q92520    Q9UHG2  \n",
       "0   731854.0  4749780.0   822784.0  42511700.0  1051980.0  322496.0  177983.0  \n",
       "1   618772.0  4223100.0   661058.0  34100500.0   738247.0  496976.0  171733.0  \n",
       "2   879126.0  5728010.0   971421.0  38324600.0   843926.0  674551.0  245188.0  \n",
       "3  1204390.0  7528540.0  1246230.0  34148900.0   816347.0  601869.0  229232.0  \n",
       "4   459165.0  8704760.0  1274950.0  45982800.0   598882.0  569973.0  176722.0  \n",
       "\n",
       "[5 rows x 28 columns]"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#We want to drop NaN values, as they may interfere with our results.\n",
    "proteins = proteins_merge.dropna(axis=1).drop(columns=['visit_month', 'patient_id'])\n",
    "proteins.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "87f4f9c5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Size of New Dataframe:  (1068, 35)\n"
     ]
    },
    {
     "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>visit_id</th>\n",
       "      <th>patient_id</th>\n",
       "      <th>visit_month</th>\n",
       "      <th>updrs_1</th>\n",
       "      <th>updrs_2</th>\n",
       "      <th>updrs_3</th>\n",
       "      <th>updrs_4</th>\n",
       "      <th>upd23b_clinical_state_on_medication</th>\n",
       "      <th>O15240</th>\n",
       "      <th>P01009</th>\n",
       "      <th>...</th>\n",
       "      <th>P02790</th>\n",
       "      <th>P05090</th>\n",
       "      <th>P06396</th>\n",
       "      <th>P07602</th>\n",
       "      <th>P10909</th>\n",
       "      <th>P23142</th>\n",
       "      <th>P41222</th>\n",
       "      <th>Q12805</th>\n",
       "      <th>Q92520</th>\n",
       "      <th>Q9UHG2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>55_0</td>\n",
       "      <td>55</td>\n",
       "      <td>0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>177775.0</td>\n",
       "      <td>14415900.0</td>\n",
       "      <td>...</td>\n",
       "      <td>4774060.0</td>\n",
       "      <td>9137820.0</td>\n",
       "      <td>2607410.0</td>\n",
       "      <td>1401860.0</td>\n",
       "      <td>11575700.0</td>\n",
       "      <td>1673460.0</td>\n",
       "      <td>33921600.0</td>\n",
       "      <td>1611130.0</td>\n",
       "      <td>982965.0</td>\n",
       "      <td>408698.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>55_6</td>\n",
       "      <td>55</td>\n",
       "      <td>6</td>\n",
       "      <td>8.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>34.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>165638.0</td>\n",
       "      <td>13330800.0</td>\n",
       "      <td>...</td>\n",
       "      <td>5627420.0</td>\n",
       "      <td>8411730.0</td>\n",
       "      <td>2183310.0</td>\n",
       "      <td>1321860.0</td>\n",
       "      <td>10062500.0</td>\n",
       "      <td>1644680.0</td>\n",
       "      <td>34435000.0</td>\n",
       "      <td>1483200.0</td>\n",
       "      <td>860236.0</td>\n",
       "      <td>369870.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>55_12</td>\n",
       "      <td>55</td>\n",
       "      <td>12</td>\n",
       "      <td>10.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>41.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>On</td>\n",
       "      <td>151073.0</td>\n",
       "      <td>13225500.0</td>\n",
       "      <td>...</td>\n",
       "      <td>5953560.0</td>\n",
       "      <td>8682640.0</td>\n",
       "      <td>2792590.0</td>\n",
       "      <td>1496970.0</td>\n",
       "      <td>10753700.0</td>\n",
       "      <td>1916350.0</td>\n",
       "      <td>37906600.0</td>\n",
       "      <td>1547140.0</td>\n",
       "      <td>969345.0</td>\n",
       "      <td>318553.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>55_36</td>\n",
       "      <td>55</td>\n",
       "      <td>36</td>\n",
       "      <td>17.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>51.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>On</td>\n",
       "      <td>101056.0</td>\n",
       "      <td>13855000.0</td>\n",
       "      <td>...</td>\n",
       "      <td>6672980.0</td>\n",
       "      <td>9146830.0</td>\n",
       "      <td>2871610.0</td>\n",
       "      <td>1424390.0</td>\n",
       "      <td>11943000.0</td>\n",
       "      <td>1879000.0</td>\n",
       "      <td>32773700.0</td>\n",
       "      <td>1725030.0</td>\n",
       "      <td>910437.0</td>\n",
       "      <td>377550.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>942_6</td>\n",
       "      <td>942</td>\n",
       "      <td>6</td>\n",
       "      <td>8.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>21.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>160526.0</td>\n",
       "      <td>13667800.0</td>\n",
       "      <td>...</td>\n",
       "      <td>5755200.0</td>\n",
       "      <td>5415620.0</td>\n",
       "      <td>1636600.0</td>\n",
       "      <td>1090240.0</td>\n",
       "      <td>6855660.0</td>\n",
       "      <td>1085970.0</td>\n",
       "      <td>26674400.0</td>\n",
       "      <td>637932.0</td>\n",
       "      <td>639154.0</td>\n",
       "      <td>306621.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 35 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   visit_id  patient_id  visit_month  updrs_1  updrs_2  updrs_3  updrs_4  \\\n",
       "0      55_0          55            0     10.0      6.0     15.0      NaN   \n",
       "2      55_6          55            6      8.0     10.0     34.0      NaN   \n",
       "4     55_12          55           12     10.0     10.0     41.0      0.0   \n",
       "8     55_36          55           36     17.0     18.0     51.0      0.0   \n",
       "15    942_6         942            6      8.0      2.0     21.0      NaN   \n",
       "\n",
       "   upd23b_clinical_state_on_medication    O15240      P01009  ...     P02790  \\\n",
       "0                                  NaN  177775.0  14415900.0  ...  4774060.0   \n",
       "2                                  NaN  165638.0  13330800.0  ...  5627420.0   \n",
       "4                                   On  151073.0  13225500.0  ...  5953560.0   \n",
       "8                                   On  101056.0  13855000.0  ...  6672980.0   \n",
       "15                                 NaN  160526.0  13667800.0  ...  5755200.0   \n",
       "\n",
       "       P05090     P06396     P07602      P10909     P23142      P41222  \\\n",
       "0   9137820.0  2607410.0  1401860.0  11575700.0  1673460.0  33921600.0   \n",
       "2   8411730.0  2183310.0  1321860.0  10062500.0  1644680.0  34435000.0   \n",
       "4   8682640.0  2792590.0  1496970.0  10753700.0  1916350.0  37906600.0   \n",
       "8   9146830.0  2871610.0  1424390.0  11943000.0  1879000.0  32773700.0   \n",
       "15  5415620.0  1636600.0  1090240.0   6855660.0  1085970.0  26674400.0   \n",
       "\n",
       "       Q12805    Q92520    Q9UHG2  \n",
       "0   1611130.0  982965.0  408698.0  \n",
       "2   1483200.0  860236.0  369870.0  \n",
       "4   1547140.0  969345.0  318553.0  \n",
       "8   1725030.0  910437.0  377550.0  \n",
       "15   637932.0  639154.0  306621.0  \n",
       "\n",
       "[5 rows x 35 columns]"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#The product of the proteins and Clinical dataset.\n",
    "#Placed on the right of Visit_id so not to interfere with important protein information.\n",
    "\n",
    "merged_df_data = clinical_data.merge(proteins, how='left', on='visit_id').dropna(subset=['O15240'])\n",
    "print(\"Size of New Dataframe: \",merged_df_data.shape)\n",
    "merged_df_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f7458b57",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "6fc51bd3",
   "metadata": {},
   "source": [
    "# Data Visualization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c3e07c51",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "9d3634fa",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Create a figure with one subplot.\n",
    "fig, ax = plt.subplots()\n",
    "\n",
    "# Calculate their mean values, then store the result in new dataframes.\n",
    "merged_updrs_1 = merged_df_data.groupby('visit_month')['updrs_1'].mean().reset_index()\n",
    "merged_updrs_2 = merged_df_data.groupby('visit_month')['updrs_2'].mean().reset_index()\n",
    "merged_updrs_3 = merged_df_data.groupby('visit_month')['updrs_3'].mean().reset_index()\n",
    "merged_updrs_4 = merged_df_data.groupby('visit_month')['updrs_4'].mean().reset_index()\n",
    "\n",
    "# Plot the data as a line plot.\n",
    "ax.plot(merged_updrs_1['visit_month'], merged_updrs_1['updrs_1'], label = 'UPDRS 1')\n",
    "ax.plot(merged_updrs_2['visit_month'], merged_updrs_2['updrs_2'], label = 'UPDRS 2')\n",
    "ax.plot(merged_updrs_3['visit_month'], merged_updrs_3['updrs_3'], label = 'UPDRS 3')\n",
    "ax.plot(merged_updrs_4['visit_month'], merged_updrs_4['updrs_4'], label = 'UPDRS 4')\n",
    "\n",
    "# Add a title and axis labels.\n",
    "ax.set_title('UPDRS Scores by Visit Month')\n",
    "ax.set_xlabel('Visit Month')\n",
    "ax.set_ylabel('UPDRS Score')\n",
    "\n",
    "# Add a legend.\n",
    "ax.legend()\n",
    "\n",
    "# Show the plot.\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "305e6414",
   "metadata": {},
   "source": [
    "# Evaluation \n",
    "The problem is predicting updrs_1, updrs_2, updrs_3 and updrs_4 values at a given visit_month and potential 6, 12 and 24 month visits after the current visit_month.\n",
    "\n",
    "Predictions are evaluated on symmetric mean absolute percentage error (SMAPE) + 1."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5eecea6b",
   "metadata": {},
   "source": [
    "# What is SMAPE?\n",
    "According to Anaplan...\n",
    "(https://help.anaplan.com/685ff9b2-6370-46ba-af10-679405937113-Understand-advanced-metrics)\n",
    "\n",
    "SMAPE (symmetric mean absolute percentage error) is an accuracy measure based on percentage (or relative) errors. \n",
    "\n",
    "SMAPE has both a lower bound and an upper bound. Since it's percentage-based, it's scale-independent, which means that it can be used to compare forecast performances between datasets. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "id": "06588aa2",
   "metadata": {},
   "outputs": [],
   "source": [
    "def smape(y_true, y_pred):\n",
    "    \"\"\"\n",
    "    Parameters\n",
    "    ----------\n",
    "    y_true: array-like of shape (n_samples)\n",
    "        Array of ground-truth values\n",
    "        \n",
    "    y_pred: array-like of shape (n_samples)\n",
    "        Array of prediction values\n",
    "        \n",
    "    Returns\n",
    "    -------\n",
    "    smape:\n",
    "    Symmetric mean absolute percentage error\n",
    "    \"\"\"\n",
    "    \n",
    "\n",
    "    smap = np.zeros(len(y_true))\n",
    "    \n",
    "    num = np.abs(y_true - y_pred)\n",
    "    dem = ((np.abs(y_true) + np.abs(y_pred)) / 2)\n",
    "    \n",
    "    pos_ind = (y_true != 0)|(y_pred != 0)\n",
    "    smap[pos_ind] = num[pos_ind] / dem[pos_ind]\n",
    "    \n",
    "    return 100 * np.mean(smap)\n",
    "\n",
    "def score(df, target_columns, prediction_columns):\n",
    "    \n",
    "    \"\"\"\n",
    "    Concatenate targets and prediction into a single array and calculate SMAPE + 1\n",
    "    \n",
    "    Parameters\n",
    "    ----------\n",
    "    target_columns: list of shape (4)\n",
    "        Array of target column names\n",
    "        \n",
    "    prediction_columns: list of shape (4)\n",
    "        Array of prediction column names\n",
    "        \n",
    "    Returns\n",
    "    -------\n",
    "    score:\n",
    "    Symmetric mean absolute percentage error\n",
    "    \"\"\"\n",
    "    \n",
    "    y_true = []\n",
    "    y_pred = []\n",
    "    \n",
    "    for target_column, prediction_column in zip(target_columns, prediction_columns):\n",
    "        target_idx = df[target_column].notna()\n",
    "        y_true.append(df.loc[target_idx, target_column].values + 1)\n",
    "        y_pred.append(df.loc[target_idx, prediction_column].values + 1)\n",
    "        \n",
    "    y_true = np.concatenate(y_true)\n",
    "    y_pred = np.concatenate(y_pred)\n",
    "        \n",
    "    score = smape(y_true=y_true, y_pred=y_pred)\n",
    "    \n",
    "    return score"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2129b0b9",
   "metadata": {},
   "source": [
    "# Feature Selection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "id": "ed0e61ad",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['O15240',\n",
       " 'P01009',\n",
       " 'P01011',\n",
       " 'P01023',\n",
       " 'P01024',\n",
       " 'P01042',\n",
       " 'P01834',\n",
       " 'P01876',\n",
       " 'P02647',\n",
       " 'P02649',\n",
       " 'P02749',\n",
       " 'P02751',\n",
       " 'P02765',\n",
       " 'P02766',\n",
       " 'P02768',\n",
       " 'P02774',\n",
       " 'P02787',\n",
       " 'P02790',\n",
       " 'P05090',\n",
       " 'P06396',\n",
       " 'P07602',\n",
       " 'P10909',\n",
       " 'P23142',\n",
       " 'P41222',\n",
       " 'Q12805',\n",
       " 'Q92520',\n",
       " 'Q9UHG2',\n",
       " 'visit_month']"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "FEATURES = merged_df_data.iloc[:, 8:].columns.tolist()\n",
    "FEATURES.append('visit_month')\n",
    "FEATURES"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "b4e94b2b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['updrs_1', 'updrs_2', 'updrs_3', 'updrs_4']"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target = merged_df_data.iloc[:,3:7].columns.tolist()\n",
    "target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "53839673",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['P01009', 'P01876', 'P02749', 'P02765', 'P02766', 'visit_month', 'updrs_1']\n",
      "['P01009', 'P01834', 'P01876', 'P02647', 'P02749', 'P02765', 'P02766', 'visit_month', 'updrs_2']\n",
      "['P01009', 'P01011', 'P01834', 'P01876', 'P02647', 'P02749', 'P02765', 'visit_month', 'updrs_3']\n",
      "['P02765', 'P02766', 'visit_month', 'updrs_4']\n"
     ]
    }
   ],
   "source": [
    "\n",
    "\"\"\"\n",
    "    the for loop iterates over a list called target. \n",
    "For each value of i in target, the code selects a set of columns (FEATURES plus i)\n",
    "from our dataframe merged_df_data and creates a new dataframe with those columns.\n",
    "\"\"\"\n",
    "\n",
    "\n",
    "\n",
    "for i in target:\n",
    "    # Select the feature columns and the target\n",
    "    cols = FEATURES + [i]\n",
    "    # Create a dataframe with these columns\n",
    "    data = merged_df_data[cols].copy()\n",
    "    \n",
    "    # Replace missing values with the mean of each column\n",
    "    data = data.fillna(data.mean())\n",
    "    \n",
    "    # Calculate the correlation matrix using the Pearson method\n",
    "    corr = data.corr(method='pearson')\n",
    "    \n",
    "    # Display the correlation of each feature with the target updrs_1\n",
    "    \n",
    "    # Select the features that have a positive correlation with updrs_1\n",
    "    pos_corr_feats = corr[i][corr[i] > 0].index.tolist()\n",
    "\n",
    "    # Keep only these features in the data dataframe\n",
    "    print(pos_corr_feats)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ded5cf88",
   "metadata": {},
   "source": [
    "# Data Splitting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "2d736530",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_copy = merged_df_data.copy()\n",
    "train_copy[target] = train_copy[target].fillna(0)\n",
    "\n",
    "# Split the data into 80% training and 20% validation sets\n",
    "train_data, valid_data = train_test_split(train_copy, test_size=0.05, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "d7f809b5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1014, 35)"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7e8f9829",
   "metadata": {},
   "source": [
    "# Data Preprocessing\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "df568827",
   "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>O15240</th>\n",
       "      <th>P01009</th>\n",
       "      <th>P01011</th>\n",
       "      <th>P01023</th>\n",
       "      <th>P01024</th>\n",
       "      <th>P01042</th>\n",
       "      <th>P01834</th>\n",
       "      <th>P01876</th>\n",
       "      <th>P02647</th>\n",
       "      <th>P02649</th>\n",
       "      <th>...</th>\n",
       "      <th>P05090</th>\n",
       "      <th>P06396</th>\n",
       "      <th>P07602</th>\n",
       "      <th>P10909</th>\n",
       "      <th>P23142</th>\n",
       "      <th>P41222</th>\n",
       "      <th>Q12805</th>\n",
       "      <th>Q92520</th>\n",
       "      <th>Q9UHG2</th>\n",
       "      <th>visit_month</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1704</th>\n",
       "      <td>-0.648156</td>\n",
       "      <td>-1.498802</td>\n",
       "      <td>-0.462789</td>\n",
       "      <td>-1.031696</td>\n",
       "      <td>-0.895202</td>\n",
       "      <td>-0.111628</td>\n",
       "      <td>-0.404672</td>\n",
       "      <td>0.466661</td>\n",
       "      <td>1.092011</td>\n",
       "      <td>-0.757661</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.786038</td>\n",
       "      <td>-1.448617</td>\n",
       "      <td>-0.807476</td>\n",
       "      <td>-0.352143</td>\n",
       "      <td>-0.372430</td>\n",
       "      <td>-1.245561</td>\n",
       "      <td>-0.729636</td>\n",
       "      <td>-1.157019</td>\n",
       "      <td>-0.617452</td>\n",
       "      <td>-0.906149</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>540</th>\n",
       "      <td>0.121764</td>\n",
       "      <td>-0.011088</td>\n",
       "      <td>0.098759</td>\n",
       "      <td>-0.197821</td>\n",
       "      <td>-0.287579</td>\n",
       "      <td>-0.649865</td>\n",
       "      <td>-0.092400</td>\n",
       "      <td>-0.305224</td>\n",
       "      <td>-1.157903</td>\n",
       "      <td>-1.595186</td>\n",
       "      <td>...</td>\n",
       "      <td>0.342706</td>\n",
       "      <td>-0.518929</td>\n",
       "      <td>-0.127045</td>\n",
       "      <td>-0.314962</td>\n",
       "      <td>-1.506832</td>\n",
       "      <td>-0.589219</td>\n",
       "      <td>-0.142701</td>\n",
       "      <td>0.036889</td>\n",
       "      <td>0.213777</td>\n",
       "      <td>-1.170334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2164</th>\n",
       "      <td>1.405526</td>\n",
       "      <td>-1.599809</td>\n",
       "      <td>0.318574</td>\n",
       "      <td>2.438592</td>\n",
       "      <td>0.506963</td>\n",
       "      <td>0.136400</td>\n",
       "      <td>-0.720364</td>\n",
       "      <td>0.061690</td>\n",
       "      <td>-0.030430</td>\n",
       "      <td>0.694137</td>\n",
       "      <td>...</td>\n",
       "      <td>0.985462</td>\n",
       "      <td>1.616917</td>\n",
       "      <td>1.624675</td>\n",
       "      <td>0.747934</td>\n",
       "      <td>0.379074</td>\n",
       "      <td>1.748291</td>\n",
       "      <td>1.065902</td>\n",
       "      <td>3.248951</td>\n",
       "      <td>2.264148</td>\n",
       "      <td>-0.641964</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>208</th>\n",
       "      <td>0.184413</td>\n",
       "      <td>-0.500087</td>\n",
       "      <td>-0.493583</td>\n",
       "      <td>-0.495485</td>\n",
       "      <td>0.102285</td>\n",
       "      <td>3.216231</td>\n",
       "      <td>-0.360229</td>\n",
       "      <td>0.675554</td>\n",
       "      <td>-1.015752</td>\n",
       "      <td>0.332744</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.434194</td>\n",
       "      <td>-0.114771</td>\n",
       "      <td>-0.159729</td>\n",
       "      <td>-0.293045</td>\n",
       "      <td>0.126118</td>\n",
       "      <td>-0.358987</td>\n",
       "      <td>0.099915</td>\n",
       "      <td>-0.130738</td>\n",
       "      <td>-0.242586</td>\n",
       "      <td>0.943146</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1924</th>\n",
       "      <td>2.172941</td>\n",
       "      <td>2.683276</td>\n",
       "      <td>1.003564</td>\n",
       "      <td>3.455156</td>\n",
       "      <td>0.521990</td>\n",
       "      <td>2.022866</td>\n",
       "      <td>1.296020</td>\n",
       "      <td>0.382233</td>\n",
       "      <td>1.905632</td>\n",
       "      <td>-0.085693</td>\n",
       "      <td>...</td>\n",
       "      <td>2.444975</td>\n",
       "      <td>1.442722</td>\n",
       "      <td>0.967813</td>\n",
       "      <td>1.784221</td>\n",
       "      <td>1.233837</td>\n",
       "      <td>1.253420</td>\n",
       "      <td>2.516314</td>\n",
       "      <td>1.137831</td>\n",
       "      <td>1.403822</td>\n",
       "      <td>-1.170334</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 28 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        O15240    P01009    P01011    P01023    P01024    P01042    P01834  \\\n",
       "1704 -0.648156 -1.498802 -0.462789 -1.031696 -0.895202 -0.111628 -0.404672   \n",
       "540   0.121764 -0.011088  0.098759 -0.197821 -0.287579 -0.649865 -0.092400   \n",
       "2164  1.405526 -1.599809  0.318574  2.438592  0.506963  0.136400 -0.720364   \n",
       "208   0.184413 -0.500087 -0.493583 -0.495485  0.102285  3.216231 -0.360229   \n",
       "1924  2.172941  2.683276  1.003564  3.455156  0.521990  2.022866  1.296020   \n",
       "\n",
       "        P01876    P02647    P02649  ...    P05090    P06396    P07602  \\\n",
       "1704  0.466661  1.092011 -0.757661  ... -0.786038 -1.448617 -0.807476   \n",
       "540  -0.305224 -1.157903 -1.595186  ...  0.342706 -0.518929 -0.127045   \n",
       "2164  0.061690 -0.030430  0.694137  ...  0.985462  1.616917  1.624675   \n",
       "208   0.675554 -1.015752  0.332744  ... -0.434194 -0.114771 -0.159729   \n",
       "1924  0.382233  1.905632 -0.085693  ...  2.444975  1.442722  0.967813   \n",
       "\n",
       "        P10909    P23142    P41222    Q12805    Q92520    Q9UHG2  visit_month  \n",
       "1704 -0.352143 -0.372430 -1.245561 -0.729636 -1.157019 -0.617452    -0.906149  \n",
       "540  -0.314962 -1.506832 -0.589219 -0.142701  0.036889  0.213777    -1.170334  \n",
       "2164  0.747934  0.379074  1.748291  1.065902  3.248951  2.264148    -0.641964  \n",
       "208  -0.293045  0.126118 -0.358987  0.099915 -0.130738 -0.242586     0.943146  \n",
       "1924  1.784221  1.233837  1.253420  2.516314  1.137831  1.403822    -1.170334  \n",
       "\n",
       "[5 rows x 28 columns]"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#DATA NORMALIZATION\n",
    "\n",
    "# create a scaler object\n",
    "scaler = StandardScaler()\n",
    "\n",
    "# fit the scaler to the training data\n",
    "scaler.fit(train_data[FEATURES])\n",
    "\n",
    "# transform the training data\n",
    "train_data[FEATURES] = scaler.transform(train_data[FEATURES])\n",
    "\n",
    "# transform the validation data\n",
    "valid_data[FEATURES] = scaler.transform(valid_data[FEATURES])\n",
    "valid_data[FEATURES].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "38a93786",
   "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>patient_id</th>\n",
       "      <th>visit_month</th>\n",
       "      <th>updrs_1</th>\n",
       "      <th>updrs_2</th>\n",
       "      <th>updrs_3</th>\n",
       "      <th>updrs_4</th>\n",
       "      <th>O15240</th>\n",
       "      <th>P01009</th>\n",
       "      <th>P01011</th>\n",
       "      <th>P01023</th>\n",
       "      <th>...</th>\n",
       "      <th>P02790</th>\n",
       "      <th>P05090</th>\n",
       "      <th>P06396</th>\n",
       "      <th>P07602</th>\n",
       "      <th>P10909</th>\n",
       "      <th>P23142</th>\n",
       "      <th>P41222</th>\n",
       "      <th>Q12805</th>\n",
       "      <th>Q92520</th>\n",
       "      <th>Q9UHG2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "      <td>54.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>35040.314815</td>\n",
       "      <td>0.143252</td>\n",
       "      <td>6.518519</td>\n",
       "      <td>6.037037</td>\n",
       "      <td>17.111111</td>\n",
       "      <td>0.981481</td>\n",
       "      <td>0.167411</td>\n",
       "      <td>0.011662</td>\n",
       "      <td>-0.136687</td>\n",
       "      <td>-0.108878</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.174988</td>\n",
       "      <td>-0.054702</td>\n",
       "      <td>-0.188455</td>\n",
       "      <td>0.024771</td>\n",
       "      <td>-0.064226</td>\n",
       "      <td>0.031201</td>\n",
       "      <td>0.012691</td>\n",
       "      <td>0.050526</td>\n",
       "      <td>-0.018280</td>\n",
       "      <td>0.070926</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>20006.662965</td>\n",
       "      <td>1.099017</td>\n",
       "      <td>5.483219</td>\n",
       "      <td>6.900815</td>\n",
       "      <td>16.124125</td>\n",
       "      <td>2.681106</td>\n",
       "      <td>0.987054</td>\n",
       "      <td>0.962793</td>\n",
       "      <td>0.770259</td>\n",
       "      <td>0.906681</td>\n",
       "      <td>...</td>\n",
       "      <td>0.795467</td>\n",
       "      <td>1.025953</td>\n",
       "      <td>1.008761</td>\n",
       "      <td>0.874924</td>\n",
       "      <td>0.931447</td>\n",
       "      <td>0.885149</td>\n",
       "      <td>0.992376</td>\n",
       "      <td>1.032901</td>\n",
       "      <td>0.934979</td>\n",
       "      <td>0.954368</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>4161.000000</td>\n",
       "      <td>-1.170334</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-1.362136</td>\n",
       "      <td>-1.637204</td>\n",
       "      <td>-2.572433</td>\n",
       "      <td>-1.632807</td>\n",
       "      <td>...</td>\n",
       "      <td>-2.751539</td>\n",
       "      <td>-2.088182</td>\n",
       "      <td>-2.395747</td>\n",
       "      <td>-1.431941</td>\n",
       "      <td>-1.708292</td>\n",
       "      <td>-2.058301</td>\n",
       "      <td>-1.847852</td>\n",
       "      <td>-1.753979</td>\n",
       "      <td>-1.410468</td>\n",
       "      <td>-1.567652</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>16970.250000</td>\n",
       "      <td>-0.906149</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>0.250000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-0.466360</td>\n",
       "      <td>-0.586696</td>\n",
       "      <td>-0.460061</td>\n",
       "      <td>-0.631988</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.511785</td>\n",
       "      <td>-0.716586</td>\n",
       "      <td>-0.805960</td>\n",
       "      <td>-0.569868</td>\n",
       "      <td>-0.671749</td>\n",
       "      <td>-0.379202</td>\n",
       "      <td>-0.587873</td>\n",
       "      <td>-0.640670</td>\n",
       "      <td>-0.683128</td>\n",
       "      <td>-0.584874</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>31082.000000</td>\n",
       "      <td>-0.113594</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>15.500000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.058790</td>\n",
       "      <td>-0.145508</td>\n",
       "      <td>-0.084044</td>\n",
       "      <td>-0.260048</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.144512</td>\n",
       "      <td>-0.137391</td>\n",
       "      <td>-0.386342</td>\n",
       "      <td>-0.200317</td>\n",
       "      <td>-0.311716</td>\n",
       "      <td>0.086798</td>\n",
       "      <td>-0.065334</td>\n",
       "      <td>-0.026736</td>\n",
       "      <td>-0.143635</td>\n",
       "      <td>-0.128105</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>56004.750000</td>\n",
       "      <td>0.943146</td>\n",
       "      <td>8.750000</td>\n",
       "      <td>9.750000</td>\n",
       "      <td>28.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.489456</td>\n",
       "      <td>0.519655</td>\n",
       "      <td>0.264573</td>\n",
       "      <td>0.225038</td>\n",
       "      <td>...</td>\n",
       "      <td>0.376209</td>\n",
       "      <td>0.364862</td>\n",
       "      <td>0.206373</td>\n",
       "      <td>0.530467</td>\n",
       "      <td>0.491665</td>\n",
       "      <td>0.426960</td>\n",
       "      <td>0.416973</td>\n",
       "      <td>0.477420</td>\n",
       "      <td>0.363799</td>\n",
       "      <td>0.659843</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>65043.000000</td>\n",
       "      <td>2.528256</td>\n",
       "      <td>24.000000</td>\n",
       "      <td>27.000000</td>\n",
       "      <td>67.000000</td>\n",
       "      <td>13.000000</td>\n",
       "      <td>3.113893</td>\n",
       "      <td>2.967346</td>\n",
       "      <td>1.328081</td>\n",
       "      <td>3.455156</td>\n",
       "      <td>...</td>\n",
       "      <td>1.498267</td>\n",
       "      <td>2.779226</td>\n",
       "      <td>3.296351</td>\n",
       "      <td>2.813231</td>\n",
       "      <td>2.640610</td>\n",
       "      <td>2.339708</td>\n",
       "      <td>3.281681</td>\n",
       "      <td>3.139003</td>\n",
       "      <td>3.248951</td>\n",
       "      <td>2.626835</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>8 rows × 33 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         patient_id  visit_month    updrs_1    updrs_2    updrs_3    updrs_4  \\\n",
       "count     54.000000    54.000000  54.000000  54.000000  54.000000  54.000000   \n",
       "mean   35040.314815     0.143252   6.518519   6.037037  17.111111   0.981481   \n",
       "std    20006.662965     1.099017   5.483219   6.900815  16.124125   2.681106   \n",
       "min     4161.000000    -1.170334   0.000000   0.000000   0.000000   0.000000   \n",
       "25%    16970.250000    -0.906149   2.000000   0.250000   0.000000   0.000000   \n",
       "50%    31082.000000    -0.113594   6.000000   4.000000  15.500000   0.000000   \n",
       "75%    56004.750000     0.943146   8.750000   9.750000  28.000000   0.000000   \n",
       "max    65043.000000     2.528256  24.000000  27.000000  67.000000  13.000000   \n",
       "\n",
       "          O15240     P01009     P01011     P01023  ...     P02790     P05090  \\\n",
       "count  54.000000  54.000000  54.000000  54.000000  ...  54.000000  54.000000   \n",
       "mean    0.167411   0.011662  -0.136687  -0.108878  ...  -0.174988  -0.054702   \n",
       "std     0.987054   0.962793   0.770259   0.906681  ...   0.795467   1.025953   \n",
       "min    -1.362136  -1.637204  -2.572433  -1.632807  ...  -2.751539  -2.088182   \n",
       "25%    -0.466360  -0.586696  -0.460061  -0.631988  ...  -0.511785  -0.716586   \n",
       "50%     0.058790  -0.145508  -0.084044  -0.260048  ...  -0.144512  -0.137391   \n",
       "75%     0.489456   0.519655   0.264573   0.225038  ...   0.376209   0.364862   \n",
       "max     3.113893   2.967346   1.328081   3.455156  ...   1.498267   2.779226   \n",
       "\n",
       "          P06396     P07602     P10909     P23142     P41222     Q12805  \\\n",
       "count  54.000000  54.000000  54.000000  54.000000  54.000000  54.000000   \n",
       "mean   -0.188455   0.024771  -0.064226   0.031201   0.012691   0.050526   \n",
       "std     1.008761   0.874924   0.931447   0.885149   0.992376   1.032901   \n",
       "min    -2.395747  -1.431941  -1.708292  -2.058301  -1.847852  -1.753979   \n",
       "25%    -0.805960  -0.569868  -0.671749  -0.379202  -0.587873  -0.640670   \n",
       "50%    -0.386342  -0.200317  -0.311716   0.086798  -0.065334  -0.026736   \n",
       "75%     0.206373   0.530467   0.491665   0.426960   0.416973   0.477420   \n",
       "max     3.296351   2.813231   2.640610   2.339708   3.281681   3.139003   \n",
       "\n",
       "          Q92520     Q9UHG2  \n",
       "count  54.000000  54.000000  \n",
       "mean   -0.018280   0.070926  \n",
       "std     0.934979   0.954368  \n",
       "min    -1.410468  -1.567652  \n",
       "25%    -0.683128  -0.584874  \n",
       "50%    -0.143635  -0.128105  \n",
       "75%     0.363799   0.659843  \n",
       "max     3.248951   2.626835  \n",
       "\n",
       "[8 rows x 33 columns]"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "valid_data.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "790a88e4",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "9f688e26",
   "metadata": {},
   "source": [
    "Now that we have our data ready for testing there are three models I will be trying out.\n",
    "\n",
    "One thing to note is that these implementations only do not utilize feature engineering for optimal performance. "
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6e3f768c",
   "metadata": {},
   "source": [
    "# Linear Regression\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a9fb96ad",
   "metadata": {},
   "source": [
    "Code below trains and evaluates a linear regression model for each target feature in the AMP Parkinson's dataset. The code defines a function called get_predictions that takes the my_train dataset and a dictionary of model objects. The my_train dataset is a Pandas DataFrame containing the training data, and the model dictionary contains one trained linear regression model for each target feature."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "id": "f04f0a1f",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "model = {}\n",
    "\n",
    "for i in target:\n",
    "        \n",
    "    temp = train_data.dropna(subset=[i]) \n",
    "\n",
    "    X = temp['visit_month']\n",
    "    y = temp[i]\n",
    "        \n",
    "    trained = LinearRegression().fit(X.values.reshape(-1, 1), y)\n",
    "    \n",
    "    model[i] = trained"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "id": "4cb16de1",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_predictions(my_train, model):\n",
    "    y_pred = []\n",
    "    my_train = my_train.fillna(0)\n",
    "    \n",
    "    for u in target:\n",
    "        \n",
    "        my_train['result_' + str(u)] = 0\n",
    "        \n",
    "        X = my_train[\"visit_month\"]\n",
    "        \n",
    "        temp = my_train[FEATURES].copy()\n",
    "        y = model[u].predict(X.values.reshape(-1, 1))\n",
    "        y_pred.append(y)\n",
    "        temp['preds'] = np.ceil(y)\n",
    "        \n",
    "        my_train['result_' + str(u)] = temp['preds']\n",
    "    \n",
    "    result = pd.DataFrame()\n",
    "\n",
    "    for m in [0, 6, 12, 24]:\n",
    "        for u in [1, 2, 3, 4]:\n",
    "\n",
    "            temp = my_train[[\"visit_id\", \"result_updrs_\" + str(u)]].copy()\n",
    "            temp[\"prediction_id\"] = temp[\"visit_id\"] + \"_updrs_\" + str(u) + \"_plus_\" + str(m) + \"_months\"\n",
    "            temp[\"rating\"] = temp[\"result_updrs_\" + str(u)]\n",
    "            temp = temp [['prediction_id', 'rating']]\n",
    "\n",
    "            result = result.append(temp)            \n",
    "    result = result.drop_duplicates(subset=['prediction_id', 'rating'])\n",
    "\n",
    "    return result,y_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "id": "a8bc8da8",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      ":SMAPE = 71.41016444863531\n",
      ":SMAPE = 105.90901556127494\n",
      ":SMAPE = 98.64918982735661\n",
      ":SMAPE = 179.47542263332096\n",
      "--> Overall results for out predictions\n",
      ": SMAPE = 113.86094811764696\n"
     ]
    }
   ],
   "source": [
    "df,y=get_predictions(valid_data, model)\n",
    "smape_scores = []\n",
    "i=0\n",
    "for u in target:\n",
    "    local_smape = smape(valid_data[u],y[i])\n",
    "    smape_scores.append(local_smape)\n",
    "    i=i+1\n",
    "    print(\":SMAPE = {}\".format(local_smape))\n",
    "smape_baseline = np.mean(smape_scores)\n",
    "print(\"--> Overall results for out predictions\")\n",
    "print(\": SMAPE = {}\".format(smape_baseline))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0fed9af0",
   "metadata": {},
   "source": [
    "# Multiple Linear Regression "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "id": "9dae372f",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_predictions2(my_train, model):\n",
    "    y_pred = []\n",
    "    my_train = my_train.fillna(0)\n",
    "    \n",
    "    for u in target:\n",
    "        \n",
    "        my_train['result_' + str(u)] = 0\n",
    "        \n",
    "        X = my_train[FEATURES]\n",
    "        \n",
    "        temp = my_train[FEATURES].copy()\n",
    "        y = model[u].predict(X.values.reshape(-1, len(FEATURES)))\n",
    "        y_pred.append(y)\n",
    "        temp['preds'] = np.ceil(y)\n",
    "        \n",
    "        my_train['result_' + str(u)] = temp['preds']\n",
    "    \n",
    "    result = pd.DataFrame()\n",
    "\n",
    "    for m in [0, 6, 12, 24]:\n",
    "        for u in [1, 2, 3, 4]:\n",
    "\n",
    "            temp = my_train[[\"visit_id\", \"result_updrs_\" + str(u)]].copy()\n",
    "            temp[\"prediction_id\"] = temp[\"visit_id\"] + \"_updrs_\" + str(u) + \"_plus_\" + str(m) + \"_months\"\n",
    "            temp[\"rating\"] = temp[\"result_updrs_\" + str(u)]\n",
    "            temp = temp [['prediction_id', 'rating']]\n",
    "\n",
    "            result = result.append(temp)            \n",
    "    result = result.drop_duplicates(subset=['prediction_id', 'rating'])\n",
    "\n",
    "    return result,y_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "611e14ce",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "49611883",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "c2046c31",
   "metadata": {},
   "source": [
    "# Neural Network Regression\n",
    "\n",
    "Neural networks can be very powerful models for regression tasks, particularly when dealing with complex and high-dimensional datasets. They can learn complex patterns and relationships in the data and are capable of generalizing to new data.\n",
    "\n",
    "However, neural networks can also be prone to overfitting, which occurs when the model learns the noise in the data rather than the underlying patterns. This can lead to poor performance on new data that the model has not seen before.\n",
    "\n",
    "Default median values of visit months have a flaw. Median updrs values might decrease as visit_month increases since dataset is not large enough and noisy. A baseline like that might overfit to training set and generalize poorly. Replacing median values with expanding window max values increases validation and leaderboard scores."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "id": "5a320e15",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.neural_network import MLPRegressor\n",
    "train_data3 = train_data.copy()\n",
    "\n",
    "model3 = {}\n",
    "\n",
    "for i in target:\n",
    "    temp3 = train_data3.dropna(subset=[i])\n",
    "\n",
    "    X3 = temp3[FEATURES]\n",
    "    y3 = temp3[i]\n",
    "\n",
    "    # Create a neural network regressor with 2 hidden layers of 50 neurons each\n",
    "    nn_reg = MLPRegressor(hidden_layer_sizes=(50, 50), max_iter=3000, random_state=42)\n",
    "\n",
    "    # Train the neural network regressor on the training data\n",
    "    trained3 = nn_reg.fit(X3, y3)\n",
    "\n",
    "    model3[i] = trained3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "id": "31ff3cb3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      ":SMAPE = 75.23294887807836\n",
      ":SMAPE = 108.43232900505662\n",
      ":SMAPE = 96.11214458458335\n",
      ":SMAPE = 183.5726890724776\n",
      "--> Overall results for out predictions\n",
      ": SMAPE = 115.83752788504898\n"
     ]
    }
   ],
   "source": [
    "valid_data3 = valid_data.copy()\n",
    "df,y=get_predictions2(valid_data3, model3)\n",
    "smape_scores = []\n",
    "i=0\n",
    "for u in target:\n",
    "    local_smape = smape(valid_data[u],y[i])\n",
    "    smape_scores.append(local_smape)\n",
    "    i=i+1\n",
    "    print(\":SMAPE = {}\".format(local_smape))\n",
    "smape_baseline = np.mean(smape_scores)\n",
    "print(\"--> Overall results for out predictions\")\n",
    "print(\": SMAPE = {}\".format(smape_baseline))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5b1dd834",
   "metadata": {},
   "source": [
    "# Random Forest Regressor\n",
    "\n",
    "Random Forest is an ensemble learning algorithm that combines multiple decision trees to make a prediction. Each decision tree is constructed using a random subset of the features and the data. Random Forest is known for its robustness, scalability, and high accuracy\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "id": "383fc029",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data4 = train_data.copy()\n",
    "\n",
    "model4 = {}\n",
    "\n",
    "for i in target:\n",
    "    temp4 = train_data4.dropna(subset=[i])\n",
    "\n",
    "    X4 = temp4[FEATURES]\n",
    "    y4 = temp4[i]\n",
    "    \n",
    "    trained4 = RandomForestRegressor(random_state=42).fit(X4, y4)\n",
    "\n",
    "    model4[i] = trained4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "id": "3427be49",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      ":SMAPE = 64.05149997796262\n",
      ":SMAPE = 99.8407774707326\n",
      ":SMAPE = 90.92005606591403\n",
      ":SMAPE = 175.57387545175163\n",
      "--> Overall results for out predictions\n",
      ": SMAPE = 107.59655224159022\n"
     ]
    }
   ],
   "source": [
    "valid_data4 = valid_data.copy()\n",
    "df,y=get_predictions2(valid_data4, model4)\n",
    "smape_scores = []\n",
    "i=0\n",
    "for u in target:\n",
    "    local_smape = smape(valid_data[u],y[i])\n",
    "    smape_scores.append(local_smape)\n",
    "    i=i+1\n",
    "    print(\":SMAPE = {}\".format(local_smape))\n",
    "smape_baseline = np.mean(smape_scores)\n",
    "print(\"--> Overall results for out predictions\")\n",
    "print(\": SMAPE = {}\".format(smape_baseline))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "80232bcb",
   "metadata": {},
   "source": [
    "# Further Evaluation\n",
    "\n",
    "a SMAPE score of 107-113 is not a good thing. The SMAPE (Symmetric Mean Absolute Percentage Error) is a measure of  the percentage difference between predicted and actual values. A lower SMAPE score indicates better accuracy, with 0 being a perfect prediction and 100 being a very poor prediction.\n",
    "\n",
    "A SMAPE score of 107 means that the predictions have an average percentage error of 107%, which is very high. This indicates that the model is not performing well and needs further optimization or refinement.\n",
    "\n",
    "We are going to try optimizing the process further."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "id": "c143ce93",
   "metadata": {},
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "Found input variables with inconsistent numbers of samples: [1014, 4]",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "Input \u001b[0;32mIn [114]\u001b[0m, in \u001b[0;36m<cell line: 1>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0m X_train, X_test, y_train, y_test \u001b[38;5;241m=\u001b[39m \u001b[43mtrain_test_split\u001b[49m\u001b[43m(\u001b[49m\u001b[43mX\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43my\u001b[49m\u001b[43m)\u001b[49m\n",
      "File \u001b[0;32m/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/site-packages/sklearn/model_selection/_split.py:2445\u001b[0m, in \u001b[0;36mtrain_test_split\u001b[0;34m(test_size, train_size, random_state, shuffle, stratify, *arrays)\u001b[0m\n\u001b[1;32m   2442\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m n_arrays \u001b[38;5;241m==\u001b[39m \u001b[38;5;241m0\u001b[39m:\n\u001b[1;32m   2443\u001b[0m     \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mValueError\u001b[39;00m(\u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mAt least one array required as input\u001b[39m\u001b[38;5;124m\"\u001b[39m)\n\u001b[0;32m-> 2445\u001b[0m arrays \u001b[38;5;241m=\u001b[39m \u001b[43mindexable\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43marrays\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m   2447\u001b[0m n_samples \u001b[38;5;241m=\u001b[39m _num_samples(arrays[\u001b[38;5;241m0\u001b[39m])\n\u001b[1;32m   2448\u001b[0m n_train, n_test \u001b[38;5;241m=\u001b[39m _validate_shuffle_split(\n\u001b[1;32m   2449\u001b[0m     n_samples, test_size, train_size, default_test_size\u001b[38;5;241m=\u001b[39m\u001b[38;5;241m0.25\u001b[39m\n\u001b[1;32m   2450\u001b[0m )\n",
      "File \u001b[0;32m/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/site-packages/sklearn/utils/validation.py:433\u001b[0m, in \u001b[0;36mindexable\u001b[0;34m(*iterables)\u001b[0m\n\u001b[1;32m    414\u001b[0m \u001b[38;5;124;03m\"\"\"Make arrays indexable for cross-validation.\u001b[39;00m\n\u001b[1;32m    415\u001b[0m \n\u001b[1;32m    416\u001b[0m \u001b[38;5;124;03mChecks consistent length, passes through None, and ensures that everything\u001b[39;00m\n\u001b[0;32m   (...)\u001b[0m\n\u001b[1;32m    429\u001b[0m \u001b[38;5;124;03m    sparse matrix, or dataframe) or `None`.\u001b[39;00m\n\u001b[1;32m    430\u001b[0m \u001b[38;5;124;03m\"\"\"\u001b[39;00m\n\u001b[1;32m    432\u001b[0m result \u001b[38;5;241m=\u001b[39m [_make_indexable(X) \u001b[38;5;28;01mfor\u001b[39;00m X \u001b[38;5;129;01min\u001b[39;00m iterables]\n\u001b[0;32m--> 433\u001b[0m \u001b[43mcheck_consistent_length\u001b[49m\u001b[43m(\u001b[49m\u001b[38;5;241;43m*\u001b[39;49m\u001b[43mresult\u001b[49m\u001b[43m)\u001b[49m\n\u001b[1;32m    434\u001b[0m \u001b[38;5;28;01mreturn\u001b[39;00m result\n",
      "File \u001b[0;32m/Library/Frameworks/Python.framework/Versions/3.10/lib/python3.10/site-packages/sklearn/utils/validation.py:387\u001b[0m, in \u001b[0;36mcheck_consistent_length\u001b[0;34m(*arrays)\u001b[0m\n\u001b[1;32m    385\u001b[0m uniques \u001b[38;5;241m=\u001b[39m np\u001b[38;5;241m.\u001b[39munique(lengths)\n\u001b[1;32m    386\u001b[0m \u001b[38;5;28;01mif\u001b[39;00m \u001b[38;5;28mlen\u001b[39m(uniques) \u001b[38;5;241m>\u001b[39m \u001b[38;5;241m1\u001b[39m:\n\u001b[0;32m--> 387\u001b[0m     \u001b[38;5;28;01mraise\u001b[39;00m \u001b[38;5;167;01mValueError\u001b[39;00m(\n\u001b[1;32m    388\u001b[0m         \u001b[38;5;124m\"\u001b[39m\u001b[38;5;124mFound input variables with inconsistent numbers of samples: \u001b[39m\u001b[38;5;132;01m%r\u001b[39;00m\u001b[38;5;124m\"\u001b[39m\n\u001b[1;32m    389\u001b[0m         \u001b[38;5;241m%\u001b[39m [\u001b[38;5;28mint\u001b[39m(l) \u001b[38;5;28;01mfor\u001b[39;00m l \u001b[38;5;129;01min\u001b[39;00m lengths]\n\u001b[1;32m    390\u001b[0m     )\n",
      "\u001b[0;31mValueError\u001b[0m: Found input variables with inconsistent numbers of samples: [1014, 4]"
     ]
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "835ee2ab",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}