running_models_PredictingDosageFrequencyIndex_OLD.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# standard library dependencies\n",
    "from collections import Counter\n",
    "\n",
    "# external dependencies\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "import numpy as np\n",
    "import pandas as pd \n",
    "\n",
    "# local dependencies\n",
    "from classification_preparation import (\n",
    "    load_dataset_with_engineered_features,\n",
    "    return_data_to_predict_at_month,\n",
    "    prepare_to_predict_feature,\n",
    "    split_into_training_validating_testing\n",
    ")\n",
    "\n",
    "from ensemble_classifiers import (\n",
    "    GB_classifier,\n",
    "    RF_classifier,\n",
    "    NB_classifier\n",
    ")\n",
    "\n",
    "from create_violin_plots import create_violin_plots_model_acc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "cleaned_01-2018.txt: (6493, 26)\n",
      "\n",
      "cleaned_01-2019.txt: (9707, 26)\n",
      "\n",
      "cleaned_02-2018.txt: (6733, 26)\n",
      "\n",
      "cleaned_02-2019.txt: (9923, 26)\n",
      "\n",
      "cleaned_03-2018.txt: (7015, 26)\n",
      "\n",
      "cleaned_04-2018.txt: (7328, 26)\n",
      "\n",
      "cleaned_05-2018.txt: (7678, 26)\n",
      "\n",
      "cleaned_06-2018.txt: (7944, 26)\n",
      "\n",
      "cleaned_07-2018.txt: (8209, 26)\n",
      "\n",
      "cleaned_08-2018.txt: (8443, 26)\n",
      "\n",
      "cleaned_09-2018.txt: (8652, 26)\n",
      "\n",
      "cleaned_10-2018.txt: (8940, 26)\n",
      "\n",
      "cleaned_11-2018.txt: (9218, 26)\n",
      "\n",
      "cleaned_12-2018.txt: (9440, 26)\n",
      "\n",
      "cleaned_2019-03.txt: (10171, 26)\n",
      "\n",
      "cleaned_2019-04.txt: (10518, 26)\n",
      "\n",
      "cleaned_2019-05.txt: (10829, 26)\n",
      "\n",
      "cleaned_2019-06.txt: (11124, 26)\n",
      "\n",
      "cleaned_2019-07.txt: (11388, 26)\n",
      "\n",
      "cleaned_2019-08.txt: (11668, 26)\n",
      "\n",
      "cleaned_2019-09.txt: (11986, 26)\n",
      "\n",
      "finished compiling 21 datasets\n",
      "loaded 21 datasets; adding engineered features\n",
      "added engineered features to all datasets\n"
     ]
    }
   ],
   "source": [
    "all_data = load_dataset_with_engineered_features(\n",
    "    \"data\",\n",
    "    must_have_substring=\"cleaned_\",\n",
    "    verbose=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "               DR Speciality  Payment Method #5  Payment Method #4  \\\n",
      "Patient ID                                                           \n",
      "Patient_12146            0.0                0.0                0.0   \n",
      "Patient_12147            0.0                1.0                0.0   \n",
      "Patient_12148            0.0                0.0                0.0   \n",
      "Patient_12149            0.0                0.0                0.0   \n",
      "Patient_12150            0.0                0.0                0.0   \n",
      "\n",
      "               DR PROVINCE  Dosage  DR ID  Payment Method #1  Status Group  \\\n",
      "Patient ID                                                                   \n",
      "Patient_12146          0.0     0.0    0.0                0.0           0.0   \n",
      "Patient_12147          4.0     0.0  835.0                0.0           3.0   \n",
      "Patient_12148          4.0     0.0  429.0                0.0           2.0   \n",
      "Patient_12149          0.0     0.0    0.0                0.0           0.0   \n",
      "Patient_12150          0.0     0.0    0.0                0.0           0.0   \n",
      "\n",
      "               Status Description  Frequency  ...  Day Enrollment Completed  \\\n",
      "Patient ID                                    ...                             \n",
      "Patient_12146                 0.0        0.0  ...                         0   \n",
      "Patient_12147                 6.0        0.0  ...                     43848   \n",
      "Patient_12148                19.0        0.0  ...                     43848   \n",
      "Patient_12149                 0.0        0.0  ...                         0   \n",
      "Patient_12150                 0.0        0.0  ...                         0   \n",
      "\n",
      "               State Change Day  On Drug Start Day  Last On Drug Day  \\\n",
      "Patient ID                                                             \n",
      "Patient_12146                 0                  0                 0   \n",
      "Patient_12147             43848                  0                 0   \n",
      "Patient_12148             43848                  0                 0   \n",
      "Patient_12149                 0                  0                 0   \n",
      "Patient_12150                 0                  0                 0   \n",
      "\n",
      "               Re-Engagement Day  Re-Engagement On Drug Start Day  \\\n",
      "Patient ID                                                          \n",
      "Patient_12146                  0                                0   \n",
      "Patient_12147                  0                                0   \n",
      "Patient_12148                  0                                0   \n",
      "Patient_12149                  0                                0   \n",
      "Patient_12150                  0                                0   \n",
      "\n",
      "               Enrollment Time  On Drug Time  # of Payment Methods  \\\n",
      "Patient ID                                                           \n",
      "Patient_12146                0             0                   0.0   \n",
      "Patient_12147                0             0                   1.0   \n",
      "Patient_12148                0             0                   1.0   \n",
      "Patient_12149                0             0                   0.0   \n",
      "Patient_12150                0             0                   0.0   \n",
      "\n",
      "               DosageFrequencyIndex  \n",
      "Patient ID                           \n",
      "Patient_12146                   0.0  \n",
      "Patient_12147                   0.0  \n",
      "Patient_12148                   0.0  \n",
      "Patient_12149                   0.0  \n",
      "Patient_12150                   0.0  \n",
      "\n",
      "[5 rows x 30 columns]\n"
     ]
    }
   ],
   "source": [
    "print(all_data[\"cleaned_2019-09.txt\"].tail())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████████| 18/18 [00:01<00:00, 10.87it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "11388 useable patients\n"
     ]
    }
   ],
   "source": [
    "useable_data = return_data_to_predict_at_month(\n",
    "    all_data,\n",
    "    \"cleaned_2019-08.txt\",\n",
    "    verbose=True\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['delta_DR Speciality', 'delta_Payment Method #5',\n",
       "       'delta_Payment Method #4', 'delta_DR PROVINCE', 'delta_Dosage',\n",
       "       'delta_DR ID', 'delta_Payment Method #1', 'delta_Status Group',\n",
       "       'delta_Status Description', 'delta_Frequency', 'delta_Diagnosis ID',\n",
       "       'delta_Payment Method #3', 'delta_Biologic Line of Therapy',\n",
       "       'delta_Status', 'delta_Gender ID', 'delta_Payment Method #2',\n",
       "       'delta_Patient Receiving Free Drug', 'delta_Case State',\n",
       "       'delta_Age Range', 'delta_Day Enrollment Received',\n",
       "       'delta_Day Enrollment Completed', 'delta_State Change Day',\n",
       "       'delta_On Drug Start Day', 'delta_Last On Drug Day',\n",
       "       'delta_Re-Engagement Day', 'delta_Re-Engagement On Drug Start Day',\n",
       "       'delta_Enrollment Time', 'delta_On Drug Time',\n",
       "       'delta_# of Payment Methods', 'delta_DosageFrequencyIndex',\n",
       "       'delta_Number of Months', 'DR Speciality', 'Payment Method #5',\n",
       "       'Payment Method #4', 'DR PROVINCE', 'Dosage', 'DR ID',\n",
       "       'Payment Method #1', 'Status Group', 'Status Description', 'Frequency',\n",
       "       'Diagnosis ID', 'Payment Method #3', 'Biologic Line of Therapy',\n",
       "       'Status', 'Gender ID', 'Payment Method #2',\n",
       "       'Patient Receiving Free Drug', 'Case State', 'Age Range',\n",
       "       'Day Enrollment Received', 'Day Enrollment Completed',\n",
       "       'State Change Day', 'On Drug Start Day', 'Last On Drug Day',\n",
       "       'Re-Engagement Day', 'Re-Engagement On Drug Start Day',\n",
       "       'Enrollment Time', 'On Drug Time', '# of Payment Methods',\n",
       "       'DosageFrequencyIndex'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "useable_data.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "delta_columns = [ colname for colname in useable_data.columns \n",
    "                 if \"delta_\" in colname\n",
    "                and colname != \"delta_DosageFrequencyIndex\"]\n",
    "\n",
    "useable_data = useable_data.drop(columns=delta_columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "useable_data[\"delta_DosageFrequencyIndex\"] = useable_data[\"delta_DosageFrequencyIndex\"].astype(bool).astype(int)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "features_df, feature_to_predict = prepare_to_predict_feature(\n",
    "    \"delta_DosageFrequencyIndex\",\n",
    "    useable_data,\n",
    "    verbose=True\n",
    ")\n",
    "labels = list(set(feature_to_predict.values.tolist()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dimensions of testing dataset: (1709, 30), (1709,)\n",
      "dimensions of validating dataset: (1936, 30), (1936,)\n",
      "dimensions of training dataset: (7743, 30), (7743,)\n",
      "dimensions of original dataset: (11388, 30)\n"
     ]
    }
   ],
   "source": [
    "all_splits_dict = split_into_training_validating_testing(\n",
    "    features_df,\n",
    "    feature_to_predict,\n",
    "    verbose=True\n",
    ")\n",
    "\n",
    "# keys of dictionaries in all_splits_dict\n",
    "#\"X train\",\n",
    "#\"X validate\", \n",
    "#\"X test\", \n",
    "#\"y train\", \n",
    "#\"y validate\", \n",
    "#\"y test\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "for splits_dict in all_splits_dict:\n",
    "    # result is a dict of:\n",
    "    # \"Model Confusion Matrix\"\n",
    "    # \"Model Accuracy\"\n",
    "    # \"Baseline Confusion Matrix\"\n",
    "    # \"Baseline Accuracy\"\n",
    "    # \"Classifier\"\n",
    "    result = RF_classifier(\n",
    "        splits_dict[\"X train\"],\n",
    "        splits_dict[\"y train\"],\n",
    "        splits_dict[\"X test\"],\n",
    "        splits_dict[\"y test\"],\n",
    "        labels,\n",
    "        return_baseline_cm = True\n",
    "    )\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.8864833235810415 0.676418958455237\n",
      "\n",
      "\n",
      "[[1314   35]\n",
      " [ 159  201]]\n",
      "\n",
      "\n",
      "[[1091  258]\n",
      " [ 295   65]]\n"
     ]
    }
   ],
   "source": [
    "print(result[\"Model Accuracy\"], result[\"Baseline Accuracy\"])\n",
    "print(\"\\n\")\n",
    "print(result[\"Model Confusion Matrix\"])\n",
    "print(\"\\n\")\n",
    "print(result[\"Baseline Confusion Matrix\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_splits_dict = split_into_training_validating_testing(\n",
    "    features_df,\n",
    "    feature_to_predict,\n",
    "    verbose=True,\n",
    "    use_strat_splits=True\n",
    ")\n",
    "\n",
    "# keys of dictionaries in all_splits_dict\n",
    "#\"X train\",\n",
    "#\"X validate\", \n",
    "#\"X test\", \n",
    "#\"y train\", \n",
    "#\"y validate\", \n",
    "#\"y test\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "========RF========\n",
      "0.8911644236395553 0.6770040959625512\n",
      "0.8794616734932709 0.6541837331772966\n",
      "0.8864833235810415 0.6582796957284962\n",
      "0.88765359859567 0.6588648332358105\n",
      "0.8847279110590989 0.6500877706260971\n",
      "0.8870684610883558 0.6676418958455237\n",
      "0.890579286132241 0.6377998829724986\n",
      "0.8864833235810415 0.6641310708016384\n",
      "0.8870684610883558 0.6489174956114687\n",
      "0.890579286132241 0.6623756582796957\n",
      "[[1307.7   41.3]\n",
      " [ 151.6  208.4]]\n",
      "========GB========\n",
      "0.9110590988882388 0.6734932709186658\n",
      "0.9227618490345231 0.6571094207138678\n",
      "0.8648332358104155 0.6459918080748976\n",
      "0.8630778232884728 0.6676418958455237\n",
      "0.874195435927443 0.6752486834406085\n",
      "0.9093036863662961 0.6565242832065535\n",
      "0.8894090111176126 0.6436512580456407\n",
      "0.8777062609713283 0.63662960795787\n",
      "0.8970157987126974 0.6699824458747806\n",
      "0.8291398478642481 0.647162083089526\n",
      "[[1235.6  113.4]\n",
      " [  85.1  274.9]]\n",
      "========NB========\n",
      "0.8537156231714453 0.6664716208308953\n",
      "0.8414277355178467 0.6617905207723815\n",
      "0.8578115857226448 0.6892919836161499\n",
      "0.8578115857226448 0.6535985956699825\n",
      "0.8513750731421884 0.6729081334113517\n",
      "0.8589818607372732 0.6594499707431246\n",
      "0.8589818607372732 0.665886483323581\n",
      "0.8677589233469866 0.6588648332358105\n",
      "0.8677589233469866 0.6600351082504389\n",
      "0.8660035108250439 0.6588648332358105\n",
      "[[1263.7   85.3]\n",
      " [ 157.1  202.9]]\n"
     ]
    }
   ],
   "source": [
    "num_classes = len(np.unique(feature_to_predict))\n",
    "RF_model_accs, RF_model_cm = [], np.zeros((num_classes,num_classes))\n",
    "GB_model_accs, GB_model_cm = [], np.zeros((num_classes,num_classes))\n",
    "NB_model_accs, NB_model_cm = [], np.zeros((num_classes,num_classes))\n",
    "baseline_accs = []\n",
    "baseline_model_cm = np.zeros((num_classes,num_classes))\n",
    "\n",
    "RF_outcomes = pd.DataFrame(np.zeros((10,2)), columns=[\"Model Acc\", \"Baseline Acc\"])\n",
    "GB_outcomes = pd.DataFrame(np.zeros((10,2)), columns=[\"Model Acc\", \"Baseline Acc\"])\n",
    "NB_outcomes = pd.DataFrame(np.zeros((10,2)), columns=[\"Model Acc\", \"Baseline Acc\"])\n",
    "\n",
    "\n",
    "for model, acc_list, avg_cm, name, df in zip(\n",
    "    [RF_classifier, GB_classifier, NB_classifier],\n",
    "    [RF_model_accs, GB_model_accs, NB_model_accs],\n",
    "    [RF_model_cm, GB_model_cm, NB_model_cm],\n",
    "    [\"RF\",\"GB\",\"NB\"],\n",
    "    [RF_outcomes, GB_outcomes, NB_outcomes]):\n",
    "    print(f\"========{name}========\")\n",
    "    for e, splits_dict in enumerate(all_splits_dict):\n",
    "        # result is a dict of:\n",
    "        # \"Model Confusion Matrix\"\n",
    "        # \"Model Accuracy\"\n",
    "        # \"Baseline Confusion Matrix\"\n",
    "        # \"Baseline Accuracy\"\n",
    "        # \"Classifier\"\n",
    "        result = model(\n",
    "            splits_dict[\"X train\"],\n",
    "            splits_dict[\"y train\"],\n",
    "            splits_dict[\"X test\"],\n",
    "            splits_dict[\"y test\"],\n",
    "            labels,\n",
    "            return_baseline_cm = True\n",
    "        )\n",
    "        print( result[\"Model Accuracy\"], result[\"Baseline Accuracy\"] )\n",
    "        acc_list.append( result[\"Model Accuracy\"] )\n",
    "        acc_list.append( result[\"Baseline Accuracy\"] )\n",
    "        df.loc[e, \"Model Acc\"] = result[\"Model Accuracy\"]\n",
    "        df.loc[e, \"Baseline Acc\"] = result[\"Baseline Accuracy\"]\n",
    "        #if e == 0:\n",
    "        #    avg_cm = (result[\"Model Confusion Matrix\"] / 10)\n",
    "        #else:\n",
    "            #pass\n",
    "        avg_cm += (result[\"Model Confusion Matrix\"] / 10)\n",
    "\n",
    "        #if baseline_model_cm is None:\n",
    "        #    baseline_model_cm = (result[\"Baseline Confusion Matrix\"] / 30)\n",
    "        #else: \n",
    "            #pass\n",
    "        baseline_accs.append(result[\"Baseline Accuracy\"])\n",
    "        baseline_model_cm += (result[\"Baseline Confusion Matrix\"] / 30)\n",
    "    print(avg_cm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x1008 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for e, (cm, name) in enumerate(\n",
    "    zip([baseline_model_cm, RF_model_cm, GB_model_cm, NB_model_cm], \n",
    "        [\"EDR baseline\", \"Random Forest\", \"Gradient Boosting\", \"Naive Bayes\"])):\n",
    "    cm_norm = cm / cm.sum()\n",
    "    ax = sns.heatmap(cm_norm, cmap=\"Reds\", cbar=False)\n",
    "    ax.xaxis.set_ticks_position('top')\n",
    "    ax.set_title(name)\n",
    "    plt.show()\n",
    "\n",
    "plt.bar(x=np.arange(cm.shape[0]), height=cm.sum(axis=1))\n",
    "\n",
    "create_violin_plots_model_acc(RF_outcomes.values, GB_outcomes.values, NB_outcomes.values, baseline_accs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}