LOOCV_nn.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index(['Unnamed: 0.2', 'Unnamed: 0.1', 'Unnamed: 0', 'CLINIC', 'YEAR',\n",
      "       'GENDERSP', 'Susceptible', 'MSM', 'MSMW', 'MSW', 'Oth/Unk/Missing',\n",
      "       'REGION', 'Midwest', 'Northeast', 'Southeast', 'Southwest', 'West',\n",
      "       'PREV_REGION', 'PREV_CLINIC', 'DELTA_REGION'],\n",
      "      dtype='object')\n"
     ]
    }
   ],
   "source": [
    "#%reset\n",
    "import seaborn as sb\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import sklearn as sk\n",
    "import numpy as np\n",
    "import eli5\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn import metrics\n",
    "from sklearn.model_selection import cross_validate, RandomizedSearchCV, RepeatedStratifiedKFold\n",
    "from sklearn.utils import resample\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.inspection import permutation_importance\n",
    "from scipy.stats import loguniform\n",
    "from imblearn.over_sampling import RandomOverSampler\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from eli5.sklearn import PermutationImportance\n",
    "## read data \n",
    "CIP_data = pd.read_csv(\"CIP_data_encode_prev.csv\")\n",
    "CIP_data_no_drop = pd.read_csv(\"CIP_data_encode_prev_not_dropped.csv\")\n",
    "print(CIP_data_no_drop.columns)\n",
    "hfont = {'fontname':'Helvetica'}\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ALB\n",
      "0.0728476821192053\n",
      "ATL\n",
      "0.056910569105691054\n",
      "BAL\n",
      "0.01444043321299639\n",
      "BHM\n",
      "0.010948905109489052\n",
      "CHI\n",
      "0.04054054054054054\n",
      "CIN\n",
      "0.006666666666666667\n",
      "CLE\n",
      "0.030612244897959183\n",
      "DAL\n",
      "0.060810810810810814\n",
      "DEN\n",
      "0.15699658703071673\n",
      "DTR\n",
      "0.0033333333333333335\n",
      "GRB\n",
      "0.017341040462427744\n",
      "HON\n",
      "0.35789473684210527\n",
      "HON\n",
      "0.35789473684210527\n",
      "LAX\n",
      "0.22705314009661837\n",
      "LBC\n",
      "0.2835820895522388\n",
      "LVG\n",
      "0.08666666666666667\n",
      "MIA\n",
      "0.19811320754716982\n",
      "MIN\n",
      "0.05652173913043478\n",
      "NOR\n",
      "0.10204081632653061\n",
      "NYC\n",
      "0.07547169811320754\n",
      "OKC\n",
      "0.043333333333333335\n",
      "ORA\n",
      "0.3458646616541353\n",
      "ORA\n",
      "0.3458646616541353\n",
      "PHI\n",
      "0.30333333333333334\n",
      "PHX\n",
      "0.11851851851851852\n",
      "POR\n",
      "0.27218934911242604\n",
      "SDG\n",
      "0.3511450381679389\n",
      "SDG\n",
      "0.3511450381679389\n",
      "SEA\n",
      "0.3181818181818182\n",
      "SFO\n",
      "0.44481605351170567\n",
      "SFO\n",
      "0.44481605351170567\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#### LOOCV based on clinic\n",
    "\n",
    "#get all clinics \n",
    "model_nn = MLPClassifier(solver = 'lbfgs', activation = 'tanh', max_iter = 3000 ,hidden_layer_sizes= 12, alpha =1.291549665014884, random_state=10, learning_rate = 'adaptive' )\n",
    "threshold_seq = np.linspace(0,1,101)\n",
    "\n",
    "oversample = RandomOverSampler(sampling_strategy = 0.5,random_state=42)\n",
    "years_train = [2000, 2001, 2002, 2003, 2004]\n",
    "year = [2006]\n",
    "### subset data based on year FIRST to esnure only clinics that appear in that year are used \n",
    "CIP_data_training_years = CIP_data_no_drop.loc[CIP_data_no_drop['YEAR'].isin(years_train)]\n",
    "CIP_data_testing_years = CIP_data_no_drop.loc[CIP_data_no_drop['YEAR'].isin(year)]\n",
    "\n",
    "## first plot overall with all clinics as a base of comparison \n",
    "X_train = CIP_data_training_years[['MSM','MSMW', 'Oth/Unk/Missing','Northeast', 'Southeast', 'Southwest', 'West','PREV_REGION', 'PREV_CLINIC','DELTA_REGION']]\n",
    "y_train = 1 - CIP_data_training_years['Susceptible']\n",
    "X_test = CIP_data_testing_years[['MSM','MSMW', 'Oth/Unk/Missing','Northeast', 'Southeast', 'Southwest', 'West','PREV_REGION', 'PREV_CLINIC','DELTA_REGION']]\n",
    "y_test = 1 - CIP_data_testing_years['Susceptible']\n",
    "cipro_R_prev = y_test.sum()/len(y_test)\n",
    "\n",
    "X_test, y_test = oversample.fit_resample(X_test,y_test)\n",
    "model_fit_train = model_nn.fit(X_train, y_train)\n",
    "y_predict_test = model_fit_train.predict(X_test)\n",
    "y_predict_proba = model_fit_train.predict_proba(X_test)\n",
    "get_effective_threshold_prev_all_clinics = []\n",
    "incorrectly_get_X_threshold_prev_all_clinics = [] #no bootstrapping, no 95% CI \n",
    "\n",
    "for threshold in threshold_seq:\n",
    "\n",
    "        y_predict_test = np.where(y_predict_proba[:, 1] > threshold, 1, 0)\n",
    "\n",
    "        tn_test , fp_test , fn_test , tp_test  = confusion_matrix(y_true=y_test, y_pred=y_predict_test).ravel()\n",
    "\n",
    "        sensitivity_test  = tp_test  / (tp_test   + fn_test )\n",
    "        specificity_test   = tn_test / (tn_test + fp_test )\n",
    "    \n",
    "        get_effective_threshold_prev_all_clinics.append(sensitivity_test * cipro_R_prev + (1 -  cipro_R_prev)) #q_p\n",
    "        incorrectly_get_X_threshold_prev_all_clinics.append((1 - cipro_R_prev) * (1 - specificity_test)) #c_p\n",
    "plt.plot(get_effective_threshold_prev_all_clinics, incorrectly_get_X_threshold_prev_all_clinics,color = \"black\", linewidth = 3)\n",
    "plt.plot([1, 1-cipro_R_prev], [1-cipro_R_prev, 0], color = \"#e5e5e5\", linestyle=\"--\")\n",
    "plt.plot(1, 1 - cipro_R_prev, marker='.', ls='none', ms=18, color = \"black\", label = \"Dual\")\n",
    "plt.plot((1-cipro_R_prev), 0, marker='*', ls='none', ms=14, color = \"black\", label = \"Cipro\")\n",
    "\n",
    "\n",
    "#loop over clinics\n",
    "clinics = CIP_data_testing_years[\"CLINIC\"].unique()\n",
    "\n",
    "for clinic in clinics:\n",
    "  try:\n",
    "          print(clinic)\n",
    "          train_data = CIP_data_training_years.loc[CIP_data_training_years['CLINIC'] != clinic]\n",
    "          #train data - does not have clinic\n",
    "          X_train = train_data[['MSM','MSMW', 'Oth/Unk/Missing','Northeast', 'Southeast', 'Southwest', 'West','PREV_REGION', 'PREV_CLINIC','DELTA_REGION']]\n",
    "          y_train = 1 - train_data['Susceptible']\n",
    "          X_train, y_train = oversample.fit_resample(X_train,y_train)\n",
    "\n",
    "          #test data - has clinic \n",
    "          test_data = CIP_data_testing_years.loc[CIP_data_testing_years['CLINIC'] == clinic]\n",
    "          X_test = test_data[['MSM','MSMW', 'Oth/Unk/Missing','Northeast', 'Southeast', 'Southwest', 'West','PREV_REGION', 'PREV_CLINIC','DELTA_REGION']]\n",
    "          y_test = 1 - test_data['Susceptible']\n",
    "          cipro_R = y_test.sum()/len(y_test) #get prevalence in year before radndom oversampling\n",
    "          print(cipro_R)\n",
    "          X_test, y_test = oversample.fit_resample(X_test,y_test)\n",
    "\n",
    "          # test \n",
    "          model_fit_train = model_nn.fit(X_train, y_train)\n",
    "          y_predict_test = model_fit_train.predict(X_test)\n",
    "          y_predict_proba = model_fit_train.predict_proba(X_test)\n",
    "          get_effective_threshold_region = []\n",
    "          incorrectly_get_X_threshold_region  = []\n",
    "          for threshold in threshold_seq:\n",
    "\n",
    "            y_predict_test = np.where(y_predict_proba[:, 1] > threshold, 1, 0)\n",
    "\n",
    "            tn_test , fp_test , fn_test , tp_test  = confusion_matrix(y_true=y_test, y_pred=y_predict_test).ravel()\n",
    "\n",
    "            sensitivity_test  = tp_test  / (tp_test   + fn_test )\n",
    "            specificity_test   = tn_test / (tn_test + fp_test )\n",
    "  \n",
    "            get_effective_threshold_region .append(sensitivity_test * cipro_R + (1 -  cipro_R)) #q_p\n",
    "            incorrectly_get_X_threshold_region .append((1 - cipro_R) * (1 - specificity_test)) #c_p\n",
    "            plt.plot(get_effective_threshold_region , incorrectly_get_X_threshold_region , color = \"#CFCCD1\", alpha = 0.5)\n",
    "            plt.plot(1, 1 - cipro_R, marker='.', ls='none', ms=10, color = \"gray\")\n",
    "            plt.plot((1-cipro_R), 0, marker='*', ls='none', ms=10, color = \"gray\")\n",
    "  except ValueError:\n",
    "            try: \n",
    "              oversample = RandomOverSampler(sampling_strategy = 0.5,random_state=42)\n",
    "              print(clinic)\n",
    "              train_data = CIP_data_training_years.loc[CIP_data_training_years['CLINIC'] != clinic]\n",
    "              #train data - does not have clinic\n",
    "              X_train = train_data[['MSM','MSMW', 'Oth/Unk/Missing','Northeast', 'Southeast', 'Southwest', 'West','PREV_REGION', 'PREV_CLINIC','DELTA_REGION']]\n",
    "              y_train = 1 - train_data['Susceptible']\n",
    "              X_train, y_train = oversample.fit_resample(X_train,y_train)\n",
    "\n",
    "              #test data - has clinic \n",
    "              test_data = CIP_data_testing_years.loc[CIP_data_testing_years['CLINIC'] == clinic]\n",
    "              X_test = test_data[['MSM','MSMW', 'Oth/Unk/Missing','Northeast', 'Southeast', 'Southwest', 'West','PREV_REGION', 'PREV_CLINIC','DELTA_REGION']]\n",
    "              y_test = 1 - test_data['Susceptible']\n",
    "              cipro_R = y_test.sum()/len(y_test) #get prevalence in year before radndom oversampling\n",
    "              print(cipro_R)\n",
    "              X_test, y_test = oversample.fit_resample(X_test,y_test)\n",
    "\n",
    "            # test \n",
    "              model_fit_train = model_nn.fit(X_train, y_train)\n",
    "              y_predict_test = model_fit_train.predict(X_test)\n",
    "              y_predict_proba = model_fit_train.predict_proba(X_test)\n",
    "              get_effective_threshold_region = []\n",
    "              incorrectly_get_X_threshold_region  = []\n",
    "              for threshold in threshold_seq:\n",
    "\n",
    "                y_predict_test = np.where(y_predict_proba[:, 1] > threshold, 1, 0)\n",
    "\n",
    "                tn_test , fp_test , fn_test , tp_test  = confusion_matrix(y_true=y_test, y_pred=y_predict_test).ravel()\n",
    "\n",
    "                sensitivity_test  = tp_test  / (tp_test   + fn_test )\n",
    "                specificity_test   = tn_test / (tn_test + fp_test )\n",
    "  \n",
    "                get_effective_threshold_region .append(sensitivity_test * cipro_R + (1 -  cipro_R)) #q_p\n",
    "                incorrectly_get_X_threshold_region .append((1 - cipro_R) * (1 - specificity_test)) #c_p\n",
    "                plt.plot(get_effective_threshold_region , incorrectly_get_X_threshold_region , color = \"#CFCCD1\", alpha = 0.5)\n",
    "                plt.plot(1, 1 - cipro_R, marker='.', ls='none', ms=10, color = \"gray\")\n",
    "                plt.plot((1-cipro_R), 0, marker='*', ls='none', ms=10, color = \"gray\")\n",
    "            except:\n",
    "              pass\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x\n",
      "x\n",
      "x\n",
      "x\n",
      "x\n",
      "x\n",
      "x\n",
      "x\n",
      "x\n",
      "x\n",
      "x\n",
      "x\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 2000x1000 with 6 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "test_years = [2005, 2006, 2007, 2008, 2009, 2010]\n",
    "labels = [\"A\", \"B\", \"C\", \"D\", \"E\", \"F\" ]\n",
    "oversample = RandomOverSampler(sampling_strategy = 0.5,random_state=42)\n",
    "text_for_graph = ['0', '0.1', '0.2', '0.3', '0.4', '0.5', '0.6', '0.7', '0.8', '0.9', '1']\n",
    "indices = [0,10,20,30,40,50,60,70,80,90,100]\n",
    "fig, axs = plt.subplots(2,3, figsize=(20, 10), facecolor='w', edgecolor='k', sharex = 'all', sharey = 'all')\n",
    "#fig.set_ylabel(\"Receives unnecessary treatment ($\\omega$(p))\", fontsize=16, **hfont)\n",
    "#fig.set_xlabel(r\"Receives effective treatment ($\\theta$(p))\", fontsize=16, **hfont)\n",
    "fig.subplots_adjust(hspace = .15, wspace=.1)\n",
    "axs = axs.ravel()\n",
    "i = 0\n",
    "for year in test_years: \n",
    "    years_train = np.array(range(year - 5, year))\n",
    "\n",
    "    CIP_data_training_years = CIP_data_no_drop.loc[CIP_data_no_drop['YEAR'].isin(years_train)]\n",
    "    CIP_data_testing_years = CIP_data_no_drop.loc[CIP_data_no_drop['YEAR'].isin([year])]\n",
    "\n",
    "    clinics = CIP_data_testing_years[\"CLINIC\"].unique()\n",
    "    for clinic in clinics: \n",
    "        train_data = CIP_data_training_years.loc[CIP_data_training_years['CLINIC'] != clinic]\n",
    "          #train data - does not have clinic\n",
    "        X_train = train_data[['MSM','MSMW', 'Oth/Unk/Missing','Northeast', 'Southeast', 'Southwest', 'West','PREV_REGION', 'PREV_CLINIC','DELTA_REGION']]\n",
    "        y_train = 1 - train_data['Susceptible']\n",
    "        X_train, y_train = oversample.fit_resample(X_train,y_train)\n",
    "\n",
    "          #test data - has clinic \n",
    "        test_data = CIP_data_testing_years.loc[CIP_data_testing_years['CLINIC'] == clinic]\n",
    "        X_test = test_data[['MSM','MSMW', 'Oth/Unk/Missing','Northeast', 'Southeast', 'Southwest', 'West','PREV_REGION', 'PREV_CLINIC','DELTA_REGION']]\n",
    "        y_test = 1 - test_data['Susceptible']\n",
    "        cipro_R = y_test.sum()/len(y_test)\n",
    "        if cipro_R > 0: \n",
    "          try:\n",
    "              X_test, y_test = oversample.fit_resample(X_test,y_test)\n",
    "\n",
    "          # test \n",
    "              model_fit_train = model_nn.fit(X_train, y_train)\n",
    "              y_predict_test = model_fit_train.predict(X_test)\n",
    "              y_predict_proba = model_fit_train.predict_proba(X_test)\n",
    "              get_effective_threshold_region = []\n",
    "              incorrectly_get_X_threshold_region  = []\n",
    "              for threshold in threshold_seq:\n",
    "\n",
    "                y_predict_test = np.where(y_predict_proba[:, 1] > threshold, 1, 0)\n",
    "\n",
    "                tn_test , fp_test , fn_test , tp_test  = confusion_matrix(y_true=y_test, y_pred=y_predict_test).ravel()\n",
    "\n",
    "                sensitivity_test  = tp_test  / (tp_test   + fn_test )\n",
    "                specificity_test   = tn_test / (tn_test + fp_test )\n",
    "  \n",
    "              get_effective_threshold_region.append(sensitivity_test * cipro_R + (1 -  cipro_R)) #q_p\n",
    "              incorrectly_get_X_threshold_region.append((1 - cipro_R) * (1 - specificity_test)) #c_p\n",
    "              axs[i].plot(get_effective_threshold_region, incorrectly_get_X_threshold_region,color = \"black\", linewidth = 3)\n",
    "\n",
    "          except ValueError:\n",
    "            try:\n",
    "              oversample = RandomOverSampler(sampling_strategy = 0.5,random_state=42)            \n",
    "              X_test, y_test = oversample.fit_resample(X_test,y_test)\n",
    "\n",
    "            # test \n",
    "              model_fit_train = model_nn.fit(X_train, y_train)\n",
    "              y_predict_test = model_fit_train.predict(X_test)\n",
    "              y_predict_proba = model_fit_train.predict_proba(X_test)\n",
    "              get_effective_threshold_region = []\n",
    "              incorrectly_get_X_threshold_region  = []\n",
    "              for threshold in threshold_seq:\n",
    "\n",
    "                y_predict_test = np.where(y_predict_proba[:, 1] > threshold, 1, 0)\n",
    "\n",
    "                tn_test , fp_test , fn_test , tp_test  = confusion_matrix(y_true=y_test, y_pred=y_predict_test).ravel()\n",
    "\n",
    "                sensitivity_test  = tp_test  / (tp_test   + fn_test )\n",
    "                specificity_test   = tn_test / (tn_test + fp_test )\n",
    "  \n",
    "                get_effective_threshold_region.append(sensitivity_test * cipro_R + (1 -  cipro_R)) #q_p\n",
    "                incorrectly_get_X_threshold_region.append((1 - cipro_R) * (1 - specificity_test)) #c_p\n",
    "              x = 0\n",
    "              axs[i].plot(get_effective_threshold_region, incorrectly_get_X_threshold_region,color = \"black\", linewidth = 3)\n",
    "              axs[i].plot([1, 1-cipro_R], [1-cipro_R, 0], color = \"#e5e5e5\", linestyle=\"--\")\n",
    "\n",
    "\n",
    "              if (i == 0) | (i == 3):\n",
    "                axs[i].set_ylabel(\"Receives unnecessary treatment ($\\omega$(p))\", fontsize=16, **hfont)\n",
    "              if(i == 3) | (i == 4) | (i == 5):\n",
    "                axs[i].set_xlabel(r\"Receives effective treatment ($\\theta$(p))\", fontsize=16, **hfont)\n",
    "\n",
    "              axs[i].plot(1, 1 - cipro_R, marker='.', ls='none', ms=18, color = \"black\", label = \"Dual\")\n",
    "              axs[i].plot((1-cipro_R), 0, marker='*', ls='none', ms=14, color = \"black\", label = \"Cipro\")\n",
    "              axs[i].text(axs[i].get_xlim()[0] , axs[i].get_ylim()[1] + 0.01, labels[i], fontsize = 16, **hfont)\n",
    "              axs[i].title.set_text(year)\n",
    "              if i == 0:\n",
    "                axs[i].legend(loc = \"upper left\")\n",
    "\n",
    "              for index in indices:\n",
    "                axs[i].plot(get_effective_threshold_region[index], incorrectly_get_X_threshold_region[index], marker='.', ls='none', ms=11, color = \"#b56576\")\n",
    "                if get_effective_threshold_region[index] > 0.996:\n",
    "                  axs[i].text(get_effective_threshold_region[index] + 0.002, incorrectly_get_X_threshold_region[index] + 0.001, text_for_graph[x], size = 9)\n",
    "                elif (get_effective_threshold_region[index] > 0.95) & (get_effective_threshold_region[index] < 0.996):\n",
    "                  axs[i].text(get_effective_threshold_region[index] + 0.005, incorrectly_get_X_threshold_region[index] - 0.005, text_for_graph[x], size = 9)\n",
    "                else:\n",
    "                  axs[i].text(get_effective_threshold_region[index], incorrectly_get_X_threshold_region[index] + 0.025, text_for_graph[x], size = 9)\n",
    "                x = x+1\n",
    "              i += 1   \n",
    "              #plt.plot(get_effective_threshold_region , incorrectly_get_X_threshold_region , color = \"gray\", alpha = 0.5)\n",
    "              #plt.plot(1, 1 - cipro_R, marker='.', ls='none', ms=10, color = \"gray\")\n",
    "              #plt.plot((1-cipro_R), 0, marker='*', ls='none', ms=10, color = \"gray\")\n",
    "            except ValueError:\n",
    "              print('x')\n",
    "              pass\n",
    "            \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "###### hyperparameter tuning for each model \n",
    "\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "GISP_init",
   "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": "3.10.9 (main, Jan 11 2023, 09:18:18) [Clang 14.0.6 ]"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "37b4cc77837642ee25c4eb6578aebe03e17eb3bb59efdde49edbbc888dbc418f"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}