notebooks/ensembleParamTest.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "/media/Windows/Users/white/Documents/UTD/Fall19/NLP.6320.501/Project/STS-Project\n"
     ]
    }
   ],
   "source": [
    "%cd .."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from scipy.stats import pearsonr\n",
    "\n",
    "from sts_wrldom.corpusReader import read_data\n",
    "from sts_wrldom.pawarModel import pawarFit_Predict, disambiguate_pipe\n",
    "from sts_wrldom.utils import log_frame, accuracy, get_scores, rmse, write_results\n",
    "\n",
    "from sts_wrldom.enrichPipe import preprocess_raw\n",
    "from sts_wrldom.depTFIDFModel import depFit_Predict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reading dev-set from: data/dev-set.txt\n",
      "Dev DF shape: (1209, 4)\n",
      "Dev Pairs Omitted: 0 = 1209 - 1209\n",
      "\n",
      "Reading train-set from: data/train-set.txt\n",
      "Train DF shape: (1484, 4)\n",
      "Train Pairs Omitted: 0 = 1484 - 1484\n",
      "\n"
     ]
    }
   ],
   "source": [
    "dfs = read_data([\"dev\", \"train\"])\n",
    "dev = dfs[\"dev\"]\n",
    "train = dfs[\"train\"]\n",
    "dev_train = dev.append(train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Enriching data from dataframe...\n",
      "S1 parse failures: 0\n",
      "S2 parse failures: 0\n",
      "\n",
      "Enriching data from dataframe...\n",
      "S1 parse failures: 0\n",
      "S2 parse failures: 0\n",
      "\n",
      "CPU times: user 1min 1s, sys: 1.28 s, total: 1min 2s\n",
      "Wall time: 1min 2s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "dev_docs = preprocess_raw(dfs[\"dev\"])\n",
    "train_docs = preprocess_raw(dfs[\"train\"])\n",
    "dev_train_docs = dev_docs + train_docs\n",
    "\n",
    "dev_dep_pred = depFit_Predict(dev_docs)\n",
    "train_dep_pred = depFit_Predict(train_docs)\n",
    "dev_train_dep_pred = depFit_Predict(dev_train_docs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Warming up PyWSD (takes ~10 secs)... took 7.2811970710754395 secs.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Disambiguating Dev...\n",
      "Disambiguating Train...\n",
      "CPU times: user 1min 35s, sys: 1.93 s, total: 1min 37s\n",
      "Wall time: 1min 37s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "dev_disam = disambiguate_pipe(dev, \"Dev\")\n",
    "train_disam = disambiguate_pipe(train, \"Train\") \n",
    "\n",
    "dev_pawar_pred = pawarFit_Predict(dev_disam)\n",
    "train_pawar_pred = pawarFit_Predict(train_disam)\n",
    "dev_train_pawar_pred = dev_pawar_pred + train_pawar_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "dev_embed_pred = pd.read_csv(\"embeds/dev_embed_predics.csv\", index_col=0)[\"noRound\"]\n",
    "train_embed_pred = pd.read_csv(\"embeds/train_embed_predics.csv\", index_col=0)[\"noRound\"]\n",
    "dev_train_embed_pred = dev_embed_pred.append(train_embed_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "assert len(dev_dep_pred) == len(dev_pawar_pred) == len(dev_embed_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "assert len(train_dep_pred) == len(train_pawar_pred) == len(train_embed_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "assert len(dev_train_dep_pred) == len(dev_train_pawar_pred) == len(dev_train_embed_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 10 s, sys: 120 ms, total: 10.1 s\n",
      "Wall time: 10.1 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "golds = dev[\"gold\"]\n",
    "\n",
    "all_acc = []\n",
    "ne_acc = []\n",
    "for a in np.arange(0.0, 1.1, .025):\n",
    "    for b in np.arange(0.0, 1.1, .025):\n",
    "        for c in np.arange(0.0, 1.1, .025):\n",
    "            if (a + b + c) == 1:\n",
    "                if a != 0 and b != 0 and c != 0:\n",
    "                    preds = []\n",
    "                    for dep, pawar, embed in zip(dev_dep_pred, dev_pawar_pred, dev_embed_pred):\n",
    "                        weighting = (dep * a) + (pawar * b) + (embed * c)\n",
    "                        preds.append(int(np.round(weighting)))\n",
    "\n",
    "                    acc = accuracy(preds, golds)\n",
    "                    _rmse = rmse(preds, golds)\n",
    "                    pear_corr = pearsonr(preds, golds)\n",
    "                    all_acc.append((acc, [a, b, c], _rmse, pear_corr, preds))\n",
    "                \n",
    "                if a != 0 and b != 0 and c == 0:\n",
    "                    preds = []\n",
    "                    for dep, pawar, embed in zip(dev_dep_pred, dev_pawar_pred, dev_embed_pred):\n",
    "                        weighting = (dep * a) + (pawar * b) + (embed * c)\n",
    "                        preds.append(int(np.round(weighting)))\n",
    "\n",
    "                    acc = accuracy(preds, golds)\n",
    "                    _rmse = rmse(preds, golds)\n",
    "                    pear_corr = pearsonr(preds, golds)\n",
    "                    ne_acc.append((acc, [a, b, c], _rmse, pear_corr, preds))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc:0.35401157981803144, params[a,b,c]:[0.4, 0.1, 0.5], rmse=1.0098767828925022, pear=(0.38559230130722294, 3.831984694201988e-44)\n",
      "\n",
      "acc:0.35814722911497104, params[a,b,c]:[0.35000000000000003, 0.07500000000000001, 0.5750000000000001], rmse=1.0098767828925022, pear=(0.38298844028946893, 1.591283075041212e-43)\n",
      "\n",
      "acc:0.3548387096774194, params[a,b,c]:[0.375, 0.1, 0.525], rmse=1.0127393670836666, pear=(0.3807679059426377, 5.304409892883672e-43)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "dev_res = sorted(all_acc, key=lambda tup: tup[3][0], reverse=True)\n",
    "for elem in dev_res[:3]:\n",
    "    print(f\"acc:{elem[0]}, params[a,b,c]:{elem[1]}, rmse={elem[2]}, pear={elem[3]}\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc:0.3523573200992556, params[a,b,c]:[0.9750000000000001, 0.025, 0.0], rmse=1.0476646998958088, pear=(0.35881824862935624, 4.783408936030424e-38)\n",
      "\n",
      "acc:0.3457402812241522, params[a,b,c]:[0.9500000000000001, 0.05, 0.0], rmse=1.053569300662559, pear=(0.3500985462248781, 3.493846630386299e-36)\n",
      "\n",
      "acc:0.3465674110835401, params[a,b,c]:[0.925, 0.07500000000000001, 0.0], rmse=1.0606114315655741, pear=(0.3421785100401286, 1.5338852995682415e-34)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "dev_ne_res = sorted(ne_acc, key=lambda tup: tup[3][0], reverse=True)\n",
    "for elem in dev_ne_res[:3]:\n",
    "    print(f\"acc:{elem[0]}, params[a,b,c]:{elem[1]}, rmse={elem[2]}, pear={elem[3]}\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Dev Gold stats: \n",
      "             count      mean       std  min  25%  50%  75%  max\n",
      "Gold_Label  1209.0  3.816377  1.063869  1.0  3.0  4.0  5.0  5.0\n",
      "\n",
      "Dev Best Ensemble Model Prediction stats: \n",
      "               count      mean       std  min  25%  50%  75%  max      RMSE  \\\n",
      "Predic_Label  1209.0  4.053763  0.437771  2.0  4.0  4.0  4.0  5.0  1.009877   \n",
      "\n",
      "              Accuracy  Pearson's R  Pearson's R p-val  \n",
      "Predic_Label  0.354012     0.385592       3.831985e-44  \n",
      "\n",
      "------\n",
      "\n",
      "Best Dev Prediction Metrics:\n",
      "{\n",
      "  \"micro\": {\n",
      "    \"recall\": 0.179,\n",
      "    \"precision\": 0.449,\n",
      "    \"fscore\": 0.14\n",
      "  },\n",
      "  \"1\": {\n",
      "    \"recall\": 0.0,\n",
      "    \"precision\": 0.0,\n",
      "    \"fscore\": 0.0\n",
      "  },\n",
      "  \"2\": {\n",
      "    \"recall\": 0.001,\n",
      "    \"precision\": 1.0,\n",
      "    \"fscore\": 0.003\n",
      "  },\n",
      "  \"3\": {\n",
      "    \"recall\": 0.022,\n",
      "    \"precision\": 0.195,\n",
      "    \"fscore\": 0.039\n",
      "  },\n",
      "  \"4\": {\n",
      "    \"recall\": 0.743,\n",
      "    \"precision\": 0.307,\n",
      "    \"fscore\": 0.434\n",
      "  },\n",
      "  \"5\": {\n",
      "    \"recall\": 0.13,\n",
      "    \"precision\": 0.745,\n",
      "    \"fscore\": 0.221\n",
      "  }\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "stats = pd.DataFrame(dev_res[0][4], columns=[\"Predic_Label\"]).describe()\n",
    "cols = [\"RMSE\", \"Accuracy\", \"Pearson's R\", \"Pearson's R p-val\"]\n",
    "vals = [dev_res[0][2], dev_res[0][0], dev_res[0][3][0], dev_res[0][3][1]]\n",
    "extra = pd.DataFrame(vals, index=cols, columns=[\"Predic_Label\"])\n",
    "print(f\"\\nDev Gold stats: \")\n",
    "print(pd.DataFrame(list(dev[\"gold\"]), columns=[\"Gold_Label\"]).describe().T)\n",
    "print(f\"\\nDev Best Ensemble Model Prediction stats: \")\n",
    "print(stats.append(extra).T)\n",
    "print(\"\\n------\")\n",
    "\n",
    "print(f\"\\nBest Dev Prediction Metrics:\")\n",
    "metrics = get_scores(dev_res[0][4], golds)\n",
    "print(json.dumps(metrics, indent=2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "dev[\"prediction\"] = dev_res[0][4]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 13 s, sys: 39.8 ms, total: 13.1 s\n",
      "Wall time: 13.1 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "golds = train[\"gold\"]\n",
    "\n",
    "t_all_acc = []\n",
    "t_ne_acc = []\n",
    "for a in np.arange(0.0, 1.1, .025):\n",
    "    for b in np.arange(0.0, 1.1, .025):\n",
    "        for c in np.arange(0.0, 1.1, .025):\n",
    "            if (a + b + c) == 1:\n",
    "                if a != 0 and b != 0 and c != 0:\n",
    "                    preds = []\n",
    "                    for dep, pawar, embed in zip(train_dep_pred, train_pawar_pred, train_embed_pred):\n",
    "                        weighting = (dep * a) + (pawar * b) + (embed * c)\n",
    "                        preds.append(int(np.round(weighting)))\n",
    "\n",
    "                    acc = accuracy(preds, golds)\n",
    "                    _rmse = rmse(preds, golds)\n",
    "                    pear_corr = pearsonr(preds, golds)\n",
    "                    t_all_acc.append((acc, [a, b, c], _rmse, pear_corr, preds))\n",
    "                \n",
    "                if a != 0 and b != 0 and c == 0:\n",
    "                    preds = []\n",
    "                    for dep, pawar, embed in zip(train_dep_pred, train_pawar_pred, train_embed_pred):\n",
    "                        weighting = (dep * a) + (pawar * b) + (embed * c)\n",
    "                        preds.append(int(np.round(weighting)))\n",
    "\n",
    "                    acc = accuracy(preds, golds)\n",
    "                    _rmse = rmse(preds, golds)\n",
    "                    pear_corr = pearsonr(preds, golds)\n",
    "                    t_ne_acc.append((acc, [a, b, c], _rmse, pear_corr, preds))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc:0.45754716981132076, params[a,b,c]:[0.35000000000000003, 0.025, 0.625], rmse=0.9141003855292907, pear=(0.43827612913155367, 1.0741145360804136e-70)\n",
      "\n",
      "acc:0.45619946091644203, params[a,b,c]:[0.4, 0.025, 0.5750000000000001], rmse=0.9133629103099814, pear=(0.4371485460940911, 2.6614212267322794e-70)\n",
      "\n",
      "acc:0.4582210242587601, params[a,b,c]:[0.325, 0.025, 0.65], rmse=0.9166768765256781, pear=(0.4343419457584354, 2.5090365690224063e-69)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "train_res = sorted(t_all_acc, key=lambda tup: tup[3][0], reverse=True)\n",
    "for elem in train_res[:3]:\n",
    "    print(f\"acc:{elem[0]}, params[a,b,c]:{elem[1]}, rmse={elem[2]}, pear={elem[3]}\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc:0.42857142857142855, params[a,b,c]:[0.9750000000000001, 0.025, 0.0], rmse=0.9601212861389101, pear=(0.40063343936566626, 2.5631570881044666e-58)\n",
      "\n",
      "acc:0.4326145552560647, params[a,b,c]:[0.9500000000000001, 0.05, 0.0], rmse=0.9601212861389101, pear=(0.3991005094631644, 7.586393565830306e-58)\n",
      "\n",
      "acc:0.43194070080862534, params[a,b,c]:[0.925, 0.07500000000000001, 0.0], rmse=0.9646724790900519, pear=(0.3895141259203974, 5.907316805424414e-55)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "train_ne_res = sorted(t_ne_acc, key=lambda tup: tup[3][0], reverse=True)\n",
    "for elem in train_ne_res[:3]:\n",
    "    print(f\"acc:{elem[0]}, params[a,b,c]:{elem[1]}, rmse={elem[2]}, pear={elem[3]}\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "train[\"prediction\"] = train_res[0][4]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 20.2 s, sys: 180 ms, total: 20.4 s\n",
      "Wall time: 20.4 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "golds = dev_train[\"gold\"]\n",
    "\n",
    "dt_all_acc = []\n",
    "dt_ne_acc = []\n",
    "for a in np.arange(0.0, 1.1, .025):\n",
    "    for b in np.arange(0.0, 1.1, .025):\n",
    "        for c in np.arange(0.0, 1.1, .025):\n",
    "            if (a + b + c) == 1:\n",
    "                if a != 0 and b != 0 and c != 0:\n",
    "                    preds = []\n",
    "                    for dep, pawar, embed in zip(dev_train_dep_pred, dev_train_pawar_pred, dev_train_embed_pred):\n",
    "                        weighting = (dep * a) + (pawar * b) + (embed * c)\n",
    "                        preds.append(int(np.round(weighting)))\n",
    "\n",
    "                    acc = accuracy(preds, golds)\n",
    "                    _rmse = rmse(preds, golds)\n",
    "                    pear_corr = pearsonr(preds, golds)\n",
    "                    dt_all_acc.append((acc, [a, b, c], _rmse, pear_corr, preds))\n",
    "                \n",
    "                if a != 0 and b != 0 and c == 0:\n",
    "                    preds = []\n",
    "                    for dep, pawar, embed in zip(dev_train_dep_pred, dev_train_pawar_pred, dev_train_embed_pred):\n",
    "                        weighting = (dep * a) + (pawar * b) + (embed * c)\n",
    "                        preds.append(int(np.round(weighting)))\n",
    "\n",
    "                    acc = accuracy(preds, golds)\n",
    "                    _rmse = rmse(preds, golds)\n",
    "                    pear_corr = pearsonr(preds, golds)\n",
    "                    dt_ne_acc.append((acc, [a, b, c], _rmse, pear_corr, preds))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc:0.4103230597846268, params[a,b,c]:[0.4, 0.025, 0.5750000000000001], rmse=0.956343761558088, pear=(0.41091746819750496, 3.1536791216700846e-110)\n",
      "\n",
      "acc:0.4103230597846268, params[a,b,c]:[0.375, 0.025, 0.6000000000000001], rmse=0.9571200147391051, pear=(0.4103025345518991, 7.151511868652823e-110)\n",
      "\n",
      "acc:0.4095803936130709, params[a,b,c]:[0.35000000000000003, 0.025, 0.625], rmse=0.9573139796848069, pear=(0.40978689200101615, 1.4190222421519156e-109)\n",
      "\n",
      "acc:0.41069439287040477, params[a,b,c]:[0.325, 0.025, 0.65], rmse=0.957701791724083, pear=(0.4096182584451843, 1.775000911631607e-109)\n",
      "\n",
      "acc:0.4110657259561827, params[a,b,c]:[0.35000000000000003, 0.05, 0.6000000000000001], rmse=0.9600253740921925, pear=(0.40742731600477916, 3.214377016983729e-108)\n",
      "\n",
      "acc:0.4095803936130709, params[a,b,c]:[0.375, 0.05, 0.5750000000000001], rmse=0.9602187521531914, pear=(0.4070072325109342, 5.587364887303476e-108)\n",
      "\n",
      "acc:0.40883772744151503, params[a,b,c]:[0.42500000000000004, 0.025, 0.55], rmse=0.9578956388653707, pear=(0.4059443927285607, 2.2550851615792993e-107)\n",
      "\n",
      "acc:0.409209060527293, params[a,b,c]:[0.4, 0.07500000000000001, 0.525], rmse=0.9629219673408241, pear=(0.4028842262240734, 1.2178119687595093e-105)\n",
      "\n",
      "acc:0.40698106201262535, params[a,b,c]:[0.4, 0.05, 0.55], rmse=0.9617643768994855, pear=(0.4017144463368427, 5.533803350980615e-105)\n",
      "\n",
      "acc:0.4099517266988489, params[a,b,c]:[0.30000000000000004, 0.025, 0.675], rmse=0.9619574053764717, pear=(0.40042524121127243, 2.9142879989321616e-104)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "dev_train_res = sorted(dt_all_acc, key=lambda tup: tup[3][0], reverse=True)\n",
    "for elem in dev_train_res[:10]:\n",
    "    print(f\"acc:{elem[0]}, params[a,b,c]:{elem[1]}, rmse={elem[2]}, pear={elem[3]}\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "acc:0.3965837356108429, params[a,b,c]:[0.9750000000000001, 0.025, 0.0], rmse=1.0038914258032772, pear=(0.37804967523357424, 3.0733488756539445e-92)\n",
      "\n",
      "acc:0.3980690679539547, params[a,b,c]:[0.9500000000000001, 0.05, 0.0], rmse=1.0044461130686257, pear=(0.37591614525798994, 3.8526043208095424e-91)\n",
      "\n",
      "acc:0.3976977348681768, params[a,b,c]:[0.925, 0.07500000000000001, 0.0], rmse=1.0070306138641276, pear=(0.3730369505653606, 1.1341517509755665e-89)\n",
      "\n",
      "acc:0.39249907166728554, params[a,b,c]:[0.9, 0.1, 0.0], rmse=1.0154762327131774, pear=(0.35772641004197553, 4.15479929891508e-82)\n",
      "\n",
      "acc:0.3965837356108429, params[a,b,c]:[0.875, 0.125, 0.0], rmse=1.0213102554807925, pear=(0.34862555955219804, 8.346617062248372e-78)\n",
      "\n",
      "acc:0.3973264017823988, params[a,b,c]:[0.8500000000000001, 0.15000000000000002, 0.0], rmse=1.0265687007881166, pear=(0.34214251525540096, 7.960745243827402e-75)\n",
      "\n",
      "acc:0.39955440029706646, params[a,b,c]:[0.8, 0.2, 0.0], rmse=1.0303597877630601, pear=(0.33949442795004564, 1.2522432574308954e-73)\n",
      "\n",
      "acc:0.3988117341255106, params[a,b,c]:[0.775, 0.225, 0.0], rmse=1.0332388986152166, pear=(0.33841745262611295, 3.811216594730949e-73)\n",
      "\n",
      "acc:0.39918306721128854, params[a,b,c]:[0.8250000000000001, 0.17500000000000002, 0.0], rmse=1.0303597877630601, pear=(0.33756741181376887, 9.145902867241025e-73)\n",
      "\n",
      "acc:0.3988117341255106, params[a,b,c]:[0.75, 0.25, 0.0], rmse=1.0395091532130987, pear=(0.3313513152490677, 5.070465722652716e-70)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "dev_train_ne_res = sorted(dt_ne_acc, key=lambda tup: tup[3][0], reverse=True)\n",
    "for elem in dev_train_ne_res[:10]:\n",
    "    print(f\"acc:{elem[0]}, params[a,b,c]:{elem[1]}, rmse={elem[2]}, pear={elem[3]}\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "dev_train[\"prediction\"] = dev_train_res[0][4]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "for df, name in zip([dev, train], [\"dev\", \"train\"]):\n",
    "    log_frame(df, name=name, tag=\"ensemble_predics\")\n",
    "    res = df[[\"id\", \"prediction\"]]\n",
    "    write_results(res, name, \"ensemblePredic\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "%matplotlib inline\n",
    "\n",
    "labels = [1, 2, 3, 4, 5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x216 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x216 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x216 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "for df, name in zip([dev, train, dev_train], [\"Dev-Set\", \"Train-Set\", \"Dev_Train-Set\"]):\n",
    "    cm = confusion_matrix(list(df[\"gold\"]), list(df[\"prediction\"]))\n",
    "    df_cm = pd.DataFrame(cm, index=labels, columns=labels)\n",
    "\n",
    "    f,(ax1,ax2) = plt.subplots(1,2,sharey=False, figsize=(10,3))\n",
    "    g1 = sns.heatmap(df_cm,annot=True, fmt='d', ax=ax1)\n",
    "    g1.set_ylabel('True Label')\n",
    "    g1.set_xlabel('Predicted Label')\n",
    "    g1.set_title(f'{name} Confusion Matrix')\n",
    "\n",
    "    cm_norm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "    df_cm_norm = pd.DataFrame(cm_norm, index=labels, columns=labels)\n",
    "    g2 = sns.heatmap(df_cm_norm,annot=True, vmin=0, vmax=1, ax=ax2)\n",
    "    g2.set_ylabel('True Label')\n",
    "    g2.set_xlabel('Predicted Label')\n",
    "    g2.set_title(f'{name} Normed Confusion Matrix')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.8.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}