projekt_CTG.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Klasifikacija zdravlja fetusa na temelju kardiotokograma"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Autori: Valentina Križ, Jelena Kurilić, Lucija Valentić"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ovaj jupyter notebook napravljen je za potrebe kolegija Strojno učenje na PMF-u Sveučilišta u Zagrebu. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Mentori: <br>\n",
    "&nbsp;&nbsp; - Tomislav Šmuc, prof. <br>\n",
    "&nbsp;&nbsp; - Matija Piškorec, asist. <br>\n",
    "&nbsp;&nbsp; - Tomislav Lipić, asist. <br>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "U ovom notebook-u prikazana je realizacija dijela projekta u kojemu pronalazimo najbolje klasifikatore za zdravlje fetusa. <br>\n",
    "Koristimo UCI <i>Cardiotocography dataset</i> koji je moguće preuzeti na [1]."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1. Dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Naš dataset se sastoji od 2126 primjeraka dobivenih automatskom obradom kardiotokograma koji su raspoređeni u 3 klase - Normal (N), Suspect (S) i Pathologic (P). <br>\n",
    "Svaki primjerak je opisan pomoću 21 značajke. <br>\n",
    "Eksploratornom analizom je zaključeno da se radi o medicinskom <b>nebalansiranom klasifikacijskom problemu</b>, stoga je u ovom radu naglasak na odabiru modela (i oversampling metode) koji će uspješno klasificirati patološke slučajeve. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. Učitavanje podataka"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Za treniranje modela korišten je poseban jupyter notebook u kojemo je napravljena podjela na stratificirane train i test skupove. Stratifikacijom smo osigurale da svaki podskup bude dobra reprezentacija cijelog skupa podataka, tj. osiguran je isti omjer klasa u sva tri skupa. <br>\n",
    "Prilikom eksploratorne analize utvrdile smo da postoji jaka koreliranost između značajki <i>Min</i> i <i>Width</i> (negativna koreliranost -0.89), te <i>Mean</i> i <i>Median</i> (pozitivna koreliranost 0.94). U slučaju jake koreliranosti dviju varijabli, preporuča se izbaciti jednu od njih [3]. Budući da su <i>Mean</i> i <i>Width</i> jače korelirane s ciljnom varijablom, odlučile smo izbaciti značajke <i>Min</i> i <i>Median</i>."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "#import potrebnih biblioteka\n",
    "import pandas as pd\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.model_selection import StratifiedKFold\n",
    "from imblearn.over_sampling import SMOTE, BorderlineSMOTE, ADASYN\n",
    "from sklearn.metrics import recall_score, accuracy_score, confusion_matrix\n",
    "from sklearn.svm import SVC\n",
    "import xgboost as xgb\n",
    "from sklearn.ensemble import RandomForestClassifier\n",
    "import pickle\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "#imena stupaca u CTG.xls koje ćemo koristiti\n",
    "column_names = ['LB', 'AC', 'FM', 'UC', 'DL', 'DS', 'DP', 'ASTV', 'MSTV', 'ALTV', 'MLTV', 'Width', 'Max',\n",
    "                   'Nmax', 'Nzeros', 'Mode', 'Mean', 'Variance', 'Tendency', 'NSP']\n",
    "\n",
    "#stupci koji sadrže int vrijednosti, samo za ljepši ispis\n",
    "int_columns = ['LB', 'ASTV', 'ALTV', 'Width', 'Max', 'Nmax', 'Nzeros','Mode', 'Mean', 'Variance',\n",
    "               'Tendency', 'NSP']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = pd.read_excel('CTG.xls', sheet_name = 'Data', skiprows = 1, \n",
    "                     usecols = [10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 23, 24, 25, 26, 27, 29, 30, 45]\n",
    "                     , names = column_names)\n",
    "\n",
    "#izbacujemo sve retke u kojima je barem jedna varijabla nepoznata (NaN)\n",
    "data = data.dropna()\n",
    "\n",
    "for col in int_columns:\n",
    "    data[col] = (data[col]).astype(int)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "names = ['LB', 'AC', 'FM', 'UC', 'DL', 'DS', 'DP', 'ASTV', 'MSTV', 'ALTV', 'MLTV', 'Width', 'Max',\n",
    "                   'Nmax', 'Nzeros', 'Mode', 'Mean', 'Variance', 'Tendency']\n",
    "X_train = pd.read_csv('X_train.csv', names = names, header = None)\n",
    "X_test = pd.read_csv('X_test.csv', names = names, header = None)\n",
    "y_train = pd.read_csv('y_train.csv', names = ['NSP'], header = None)\n",
    "y_test = pd.read_csv('y_test.csv', names = ['NSP'], header = None)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3. Testiranje modela"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Svi testirani modeli učitani su iz <i>pickle</i> datoteka koje su nastale prilikom treniranja u treniranje.ipynb. <br>\n",
    "Usporedile smo sljedeće algoritme strojnog učenja: SVC (Support Vector Classification), XGBoost (Extreme Gradient Boosting) i Random Forest. Za svaki navedeni algoritam dobile smo 4 različita modela - bez oversamplinga i uz 3 različite metode oversamplinga (SMOTE, BorderlineSMOTE i ADASYN). <br><br>\n",
    "Optimalni parametri modela određeni su pomoću GridSearchCV-a. Za XGBoost optimizirale smo sljedeće parametre: min_child_weight ϵ {1, 5, 10}, gamma ϵ {0.5, 1, 1.5, 2, 5}, subsample ϵ {0.6, 0.8, 1.0}, colsample_bytree ϵ {0.6, 0.8, 1.0} i max_depth ϵ {3, 4, 5}. Optimizirani parametri za Random forest su n_estimators ϵ {16, 32, 64, 128} i max_features ϵ {1, 2, 3, …, 19}. Za SVC optimizirale smo parametar c ϵ {1, 10, 100, 1000} uz linear kernel, te parametre c ϵ {1, 10, 100, 1000} i gamma ϵ {0.001, 0.0001} uz rbf kernel. <br><br>\n",
    "Za svaki od modela odredile smo točnost i preciznost (omjer broja prepoznatih patoloških primjeraka i ukupnog broja patoloških primjeraka) na test skupu. <br><br>\n",
    "Za treniranje i testiranje modela potrebno je instalirati imbalanced-learn [2] i XGBoost [4]."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<b>SVC:</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Točnost SVC bez oversamplinga: 0.9061\n",
      "Osjetljivost SVC bez oversamplinga: 0.7429\n",
      "Točnost SVC uz SMOTE: 0.8920\n",
      "Osjetljivost SVC uz SMOTE: 0.8857\n",
      "Točnost SVC uz BorderlineSMOTE: 0.8592\n",
      "Osjetljivost SVC uz BorderlineSMOTE: 0.8571\n",
      "Točnost SVC uz ADASYN: 0.8732\n",
      "Osjetljivost SVC uz ADASYN: 0.9143\n"
     ]
    }
   ],
   "source": [
    "with open('svc_SMOTE.pkl', 'rb') as file:  \n",
    "    svc_SMOTE = pickle.load(file)\n",
    "with open('svc_BorderlineSMOTE.pkl', 'rb') as file:  \n",
    "    svc_BorderlineSMOTE = pickle.load(file)\n",
    "with open('svc_ADASYN.pkl', 'rb') as file:  \n",
    "    svc_ADASYN = pickle.load(file)\n",
    "with open('svc.pkl', 'rb') as file:  \n",
    "    svc_obicni = pickle.load(file)\n",
    "\n",
    "print(\"Točnost SVC bez oversamplinga: %.4f\" %svc_obicni.score(X_test, y_test))\n",
    "print(\"Osjetljivost SVC bez oversamplinga: %.4f\" %recall_score(y_test, svc_obicni.predict(X_test), average = None)[2])\n",
    "\n",
    "print(\"Točnost SVC uz SMOTE: %.4f\" %svc_SMOTE.score(X_test, y_test))\n",
    "print(\"Osjetljivost SVC uz SMOTE: %.4f\" %recall_score(y_test, svc_SMOTE.predict(X_test), average = None)[2])\n",
    "\n",
    "print(\"Točnost SVC uz BorderlineSMOTE: %.4f\" %svc_BorderlineSMOTE.score(X_test, y_test))\n",
    "print(\"Osjetljivost SVC uz BorderlineSMOTE: %.4f\" %recall_score(y_test, svc_BorderlineSMOTE.predict(X_test), average = None)[2])\n",
    "\n",
    "print(\"Točnost SVC uz ADASYN: %.4f\" %svc_ADASYN.score(X_test, y_test))\n",
    "print(\"Osjetljivost SVC uz ADASYN: %.4f\" %recall_score(y_test, svc_ADASYN.predict(X_test), average = None)[2])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<b>XGBoost:</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Točnost xgBoost bez oversamplinga: 0.9343\n",
      "Osjetljivost xgBoost bez oversamplinga: 0.9143\n",
      "Točnost xgBoost uz SMOTE: 0.9319\n",
      "Osjetljivost xgBoost uz SMOTE: 0.9143\n",
      "Točnost xgBoost uz BorderlineSMOTE: 0.9413\n",
      "Osjetljivost xgBoost uz BorderlineSMOTE: 0.9143\n",
      "Točnost xgBoost uz ADASYN: 0.9366\n",
      "Osjetljivost xgBoost uz ADASYN: 0.9429\n"
     ]
    }
   ],
   "source": [
    "with open('xgb_SMOTE.pkl', 'rb') as file:  \n",
    "    xgb_SMOTE = pickle.load(file)\n",
    "with open('xgb_BorderlineSMOTE.pkl', 'rb') as file:  \n",
    "    xgb_BorderlineSMOTE = pickle.load(file)\n",
    "with open('xgb_ADASYN.pkl', 'rb') as file:  \n",
    "    xgb_ADASYN = pickle.load(file)\n",
    "with open('xgb.pkl', 'rb') as file:  \n",
    "    xgb_obicni = pickle.load(file)\n",
    "\n",
    "print(\"Točnost xgBoost bez oversamplinga: %.4f\" %xgb_obicni.score(X_test.values, y_test.values))\n",
    "print(\"Osjetljivost xgBoost bez oversamplinga: %.4f\" %recall_score(y_test.values, xgb_obicni.predict(X_test.values), average = None)[2])\n",
    "\n",
    "print(\"Točnost xgBoost uz SMOTE: %.4f\" %xgb_SMOTE.score(X_test.values, y_test.values))\n",
    "print(\"Osjetljivost xgBoost uz SMOTE: %.4f\" %recall_score(y_test.values, xgb_SMOTE.predict(X_test.values), average = None)[2])\n",
    "\n",
    "print(\"Točnost xgBoost uz BorderlineSMOTE: %.4f\" %xgb_BorderlineSMOTE.score(X_test.values, y_test.values))\n",
    "print(\"Osjetljivost xgBoost uz BorderlineSMOTE: %.4f\" %recall_score(y_test.values, xgb_BorderlineSMOTE.predict(X_test.values), average = None)[2])\n",
    "\n",
    "print(\"Točnost xgBoost uz ADASYN: %.4f\" %xgb_ADASYN.score(X_test.values, y_test.values))\n",
    "print(\"Osjetljivost xgBoost uz ADASYN: %.4f\" %recall_score(y_test.values, xgb_ADASYN.predict(X_test.values), average = None)[2])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<b>Random Forest:</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Točnost Random forest bez oversamplinga: 0.9249\n",
      "Osjetljivost Random forest bez oversamplinga: 0.8571\n",
      "Točnost Random forest uz SMOTE: 0.9225\n",
      "Osjetljivost Random forest uz SMOTE: 0.8571\n",
      "Točnost Random forest uz BorderlineSMOTE: 0.9319\n",
      "Osjetljivost Random forest uz BorderlineSMOTE: 0.9143\n",
      "Točnost Random forest uz ADASYN: 0.9225\n",
      "Osjetljivost Random forest uz ADASYN: 0.9143\n"
     ]
    }
   ],
   "source": [
    "with open('rf_SMOTE.pkl', 'rb') as file:  \n",
    "    rf_SMOTE = pickle.load(file)\n",
    "with open('rf_BorderlineSMOTE.pkl', 'rb') as file:  \n",
    "    rf_BorderlineSMOTE = pickle.load(file)\n",
    "with open('rf_ADASYN.pkl', 'rb') as file:  \n",
    "    rf_ADASYN = pickle.load(file)\n",
    "with open('rf.pkl', 'rb') as file:  \n",
    "    rf_obicni = pickle.load(file)\n",
    "\n",
    "print(\"Točnost Random forest bez oversamplinga: %.4f\" %rf_obicni.score(X_test, y_test))\n",
    "print(\"Osjetljivost Random forest bez oversamplinga: %.4f\" %recall_score(y_test, rf_obicni.predict(X_test), average = None)[2])\n",
    "\n",
    "print(\"Točnost Random forest uz SMOTE: %.4f\" %rf_SMOTE.score(X_test, y_test))\n",
    "print(\"Osjetljivost Random forest uz SMOTE: %.4f\" %recall_score(y_test, rf_SMOTE.predict(X_test), average = None)[2])\n",
    "\n",
    "print(\"Točnost Random forest uz BorderlineSMOTE: %.4f\" %rf_BorderlineSMOTE.score(X_test, y_test))\n",
    "print(\"Osjetljivost Random forest uz BorderlineSMOTE: %.4f\" %recall_score(y_test, rf_BorderlineSMOTE.predict(X_test), average = None)[2])\n",
    "\n",
    "print(\"Točnost Random forest uz ADASYN: %.4f\" %rf_ADASYN.score(X_test, y_test))\n",
    "print(\"Osjetljivost Random forest uz ADASYN: %.4f\" %recall_score(y_test, rf_ADASYN.predict(X_test), average = None)[2])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4. Prikaz rezultata"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Za svaki od algoritama prikazale smo konfuzijske matrice koje predstavljaju rezultate dobivene modelom bez oversamplinga, te modelom uz oversampling metodu koja je dala najbolje rezultate za taj algoritam. <br>Konkretno, prikazale smo konfuzijske matrice za: <br>\n",
    "&nbsp; &nbsp; - SVC bez oversamplinga i SVC uz ADASYN oversampling <br>\n",
    "&nbsp; &nbsp; - XGBoost bez oversamplinga i XGBoost uz ADASYN oversampling <br>\n",
    "&nbsp; &nbsp; - Random Forest bez oversamplinga i Random Forest uz BorderlineSMOTE oversampling <br><br>\n",
    "Konačno, odredile smo najvažnije značajke u XGBoost modelu uz ADASYN i Random Forest modelu uz BorderlineSMOTE. Dobivene značajke pokazale su se bitnima za zdravlje fetusa i u mnogim medicinskim radovima."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<b>SVC:</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x288 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=[10, 4])\n",
    "fig.subplots_adjust(wspace=0.5, hspace=0.2)\n",
    "ax = fig.add_subplot(111)  \n",
    "ax1 = fig.add_subplot(121)\n",
    "ax2 = fig.add_subplot(122)\n",
    "\n",
    "ax.spines['top'].set_color('none')\n",
    "ax.spines['bottom'].set_color('none')\n",
    "ax.spines['left'].set_color('none')\n",
    "ax.spines['right'].set_color('none')\n",
    "ax.tick_params(labelcolor='w', top=False, bottom=False, left=False, right=False)\n",
    "ax.grid(False)\n",
    "ax.set_facecolor('white')\n",
    "\n",
    "svc_cm = confusion_matrix(y_test, svc_obicni.predict(X_test))\n",
    "svc_ADASYN_cm = confusion_matrix(y_test, svc_ADASYN.predict(X_test))\n",
    "\n",
    "label = [\"N\",\"S\",\"P\"]\n",
    "sns.set(font_scale=1.0)\n",
    "ax1.set_title('SVC', fontsize = 14)\n",
    "ax2.set_title('SVC ADASYN', fontsize = 14)\n",
    "ax.set_xlabel('Stvarna vrijednost')\n",
    "ax.set_ylabel('Predviđeno modelom')\n",
    "\n",
    "sns.heatmap(svc_cm, annot=True, xticklabels=label, yticklabels=label, cmap=\"YlGnBu\", ax=ax1)\n",
    "sns.heatmap(svc_ADASYN_cm, annot=True, xticklabels=label, yticklabels=label, cmap=\"YlGnBu\", ax=ax2)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<b>XGBoost:</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x288 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=[10, 4])\n",
    "fig.subplots_adjust(wspace=0.5, hspace=0.2)\n",
    "ax = fig.add_subplot(111)  \n",
    "ax1 = fig.add_subplot(121)\n",
    "ax2 = fig.add_subplot(122)\n",
    "\n",
    "xgb_cm = confusion_matrix(y_test.values, xgb_obicni.predict(X_test.values))\n",
    "xgb_ADASYN_cm = confusion_matrix(y_test.values, xgb_ADASYN.predict(X_test.values))\n",
    "\n",
    "ax.spines['top'].set_color('none')\n",
    "ax.spines['bottom'].set_color('none')\n",
    "ax.spines['left'].set_color('none')\n",
    "ax.spines['right'].set_color('none')\n",
    "ax.tick_params(labelcolor='w', top=False, bottom=False, left=False, right=False)\n",
    "ax.grid(False)\n",
    "ax.set_facecolor('white')\n",
    "\n",
    "ax1.set_title('XGB', fontsize = 14)\n",
    "ax2.set_title('XGB ADASYN', fontsize = 14)\n",
    "label = [\"N\",\"S\",\"P\"]\n",
    "ax.set_xlabel('Stvarna vrijednost')\n",
    "ax.set_ylabel('Predviđeno modelom')\n",
    "sns.heatmap(xgb_cm, annot=True, xticklabels=label, yticklabels=label, cmap=\"YlGnBu\", ax=ax1)\n",
    "sns.heatmap(xgb_ADASYN_cm, annot=True, xticklabels=label, yticklabels=label, cmap=\"YlGnBu\", ax=ax2)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<b>Random forest:</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x288 with 5 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=[10, 4])  \n",
    "fig.subplots_adjust(wspace=0.5, hspace=0.2)\n",
    "ax = fig.add_subplot(111) \n",
    "ax1 = fig.add_subplot(121)\n",
    "ax2 = fig.add_subplot(122)\n",
    "\n",
    "rf_cm = confusion_matrix(y_test, rf_obicni.predict(X_test))\n",
    "rf_BorderlineSMOTE_cm = confusion_matrix(y_test, rf_BorderlineSMOTE.predict(X_test))\n",
    "\n",
    "ax.spines['top'].set_color('none')\n",
    "ax.spines['bottom'].set_color('none')\n",
    "ax.spines['left'].set_color('none')\n",
    "ax.spines['right'].set_color('none')\n",
    "ax.tick_params(labelcolor='w', top=False, bottom=False, left=False, right=False)\n",
    "ax.grid(False)\n",
    "ax.set_facecolor('white')\n",
    "\n",
    "ax1.set_title('RF', fontsize = 14)\n",
    "ax2.set_title('RF BorderlineSMOTE', fontsize = 14)\n",
    "label = [\"N\",\"S\",\"P\"]\n",
    "ax.set_xlabel('Stvarna vrijednost')\n",
    "ax.set_ylabel('Predviđeno modelom')\n",
    "sns.heatmap(rf_cm, annot=True, xticklabels=label, yticklabels=label, cmap=\"YlGnBu\", ax=ax1)\n",
    "sns.heatmap(rf_BorderlineSMOTE_cm, annot=True, xticklabels=label, yticklabels=label, cmap=\"YlGnBu\", ax=ax2)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 720x360 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(10,5))\n",
    "importances = {}\n",
    "imp = rf_BorderlineSMOTE.named_steps[\"classification\"].feature_importances_\n",
    "for i in range(19):\n",
    "    importances[column_names[i]] = imp[i]\n",
    "fig = plt.figure(figsize=(10,5))\n",
    "plt.bar(range(len(importances)), list(importances.values()), align='center')\n",
    "plt.xticks(range(len(importances)), list(importances.keys()), rotation='vertical')\n",
    "plt.xlabel('Obilježje')\n",
    "plt.ylabel('Važnost obilježja')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 720x360 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(10,5))\n",
    "importances = {}\n",
    "imp = xgb_ADASYN.named_steps[\"classification\"].feature_importances_\n",
    "for i in range(19):\n",
    "    importances[column_names[i]] = imp[i]\n",
    "fig = plt.figure(figsize=(10,5))\n",
    "plt.bar(range(len(importances)), list(importances.values()), align='center')\n",
    "plt.xticks(range(len(importances)), list(importances.keys()), rotation='vertical')\n",
    "plt.xlabel('Obilježje')\n",
    "plt.ylabel('Važnost obilježja')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Literatura"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "[1] https://archive.ics.uci.edu/ml/datasets/cardiotocography <br>\n",
    "[2] https://imbalanced-learn.readthedocs.io/en/stable/ <br>\n",
    "[3] https://towardsdatascience.com/feature-selection-correlation-and-p-value-da8921bfb3cf <br>\n",
    "[4] https://xgboost.readthedocs.io/en/latest/"
   ]
  },
  {
   "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}