Final_HF_Classification.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "HF_Classification_and_Feature_Importance_Exploration.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "authorship_tag": "ABX9TyOYi6k2u4B2HnNmtGQPO5fz",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/JARACH-209/AI-ML_Lab/blob/master/Final_HF_Classification.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-QdZYxe0uIXN",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "41f8cdbf-8454-4baa-eb22-39b6640e5747"
      },
      "source": [
        "!pip install catboost\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "from sklearn import preprocessing\n",
        "from sklearn.ensemble import RandomForestClassifier\n",
        "from sklearn.neighbors import KNeighborsClassifier\n",
        "from sklearn.model_selection import cross_validate\n",
        "from xgboost import XGBClassifier\n",
        "from xgboost import XGBRFClassifier\n",
        "from lightgbm import LGBMClassifier\n",
        "from catboost import CatBoostClassifier\n",
        "from sklearn.metrics import accuracy_score\n",
        "from sklearn.metrics import roc_auc_score\n",
        "from sklearn.metrics import recall_score\n",
        "from sklearn.metrics import f1_score\n",
        "from sklearn.metrics import precision_score\n",
        "from sklearn.model_selection import train_test_split\n",
        "import seaborn as sns\n",
        "plt.style.use(\"ggplot\")\n",
        "plt.style.use(\"seaborn\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Collecting catboost\n",
            "\u001b[?25l  Downloading https://files.pythonhosted.org/packages/20/37/bc4e0ddc30c07a96482abf1de7ed1ca54e59bba2026a33bca6d2ef286e5b/catboost-0.24.4-cp36-none-manylinux1_x86_64.whl (65.7MB)\n",
            "\u001b[K     |████████████████████████████████| 65.8MB 42kB/s \n",
            "\u001b[?25hRequirement already satisfied: scipy in /usr/local/lib/python3.6/dist-packages (from catboost) (1.4.1)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from catboost) (1.15.0)\n",
            "Requirement already satisfied: numpy>=1.16.0 in /usr/local/lib/python3.6/dist-packages (from catboost) (1.19.4)\n",
            "Requirement already satisfied: matplotlib in /usr/local/lib/python3.6/dist-packages (from catboost) (3.2.2)\n",
            "Requirement already satisfied: pandas>=0.24.0 in /usr/local/lib/python3.6/dist-packages (from catboost) (1.1.5)\n",
            "Requirement already satisfied: plotly in /usr/local/lib/python3.6/dist-packages (from catboost) (4.4.1)\n",
            "Requirement already satisfied: graphviz in /usr/local/lib/python3.6/dist-packages (from catboost) (0.10.1)\n",
            "Requirement already satisfied: kiwisolver>=1.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (1.3.1)\n",
            "Requirement already satisfied: python-dateutil>=2.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (2.8.1)\n",
            "Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (0.10.0)\n",
            "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.1 in /usr/local/lib/python3.6/dist-packages (from matplotlib->catboost) (2.4.7)\n",
            "Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.6/dist-packages (from pandas>=0.24.0->catboost) (2018.9)\n",
            "Requirement already satisfied: retrying>=1.3.3 in /usr/local/lib/python3.6/dist-packages (from plotly->catboost) (1.3.3)\n",
            "Installing collected packages: catboost\n",
            "Successfully installed catboost-0.24.4\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "eUi_4XNWoDs6"
      },
      "source": [
        "# Function to run 10 fold cross validation\n",
        "def cross_validation(clf,X,Y):\n",
        "    cv_rf = cross_validate(clf,X,Y,scoring=('accuracy','roc_auc','recall','precision'),cv = 10,n_jobs=-1)\n",
        "    print(\"Average CV Accuracy Test \\t%0.2f\"%(cv_rf['test_accuracy'].mean()*100))\n",
        "    print(\"Average CV ROC-AUC Score \\t%0.2f\"%(cv_rf['test_roc_auc'].mean()*100))\n",
        "    print(\"Average CV Recall Score \\t%0.2f\"%(cv_rf['test_recall'].mean()*100))\n",
        "    print(\"Average CV Precision Score \\t%0.2f\"%(cv_rf['test_precision'].mean()*100))\n",
        "    print(\"-----------------------------\")\n",
        "    print(\"Best Accuracy CV : \\t\\t%0.2f\"%max(cv_rf['test_accuracy']*100))\n",
        "    print(\"Best ROC CV : \\t\\t\\t%0.2f\"%max(cv_rf['test_roc_auc']*100))\n",
        "    print(\"Best Recall CV : \\t\\t%0.2f\"%max(cv_rf['test_recall']*100))\n",
        "    print(\"Best Precision CV : \\t\\t%0.2f\"%max(cv_rf['test_precision']*100))\n",
        "\n",
        "def scaling_X(X_fit,X_transform):\n",
        "    sscaler = preprocessing.StandardScaler().fit(X_fit)\n",
        "    X = sscaler.transform(X_transform)\n",
        "    return X"
      ],
      "execution_count": 58,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ikkyxT4KctT4"
      },
      "source": [
        "df = pd.read_csv(\"/content/lvef_classification_DataFrame.csv\")\n",
        "df.drop(columns=['Unnamed: 0'],inplace=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "N-gi1AutIdD5"
      },
      "source": [
        "features = {}\n",
        "columns = df.columns\n",
        "for i in range(len(columns)):\n",
        "    features[i] = columns[i]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0MDyKsZEw_c4"
      },
      "source": [
        "X = df.iloc[:,:25].to_numpy()\n",
        "Y = df['lvef_classification'].to_numpy()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qZvcauNYwg7X"
      },
      "source": [
        "Y = Y.astype(int)\n",
        "Y = Y.reshape(-1)\n",
        "x_train, x_test, y_train, y_test = train_test_split(X,Y,test_size = 0.148)\n",
        "scaler = preprocessing.StandardScaler().fit(x_train)\n",
        "x_train = scaler.transform(x_train)\n",
        "scaler = preprocessing.StandardScaler().fit(x_test)\n",
        "x_test = scaler.transform(x_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rLMFh6JbzoGG",
        "outputId": "c7718243-81aa-4d71-9732-757da617dd14"
      },
      "source": [
        "clf = RandomForestClassifier(n_estimators=1333,bootstrap=True,max_depth=80, max_features='sqrt', min_samples_split= 4, min_samples_leaf=3).fit(x_train,y_train)\n",
        "clf.fit(x_train,y_train)\n",
        "y_predicted = clf.predict(x_test)\n",
        "y_train_pred = clf.predict(x_train)\n",
        "\n",
        "pred_prob = clf.predict_proba(x_test)\n",
        "print(\"Accuracy Test %0.2f\" %(accuracy_score(y_test,y_predicted)*100))\n",
        "print(\"Accuracy Train %0.2f\" %(accuracy_score(y_train,y_train_pred)*100))\n",
        "print(\"ROC-AUC Score %0.2f\" %(roc_auc_score(y_test,pred_prob[:,1])*100))\n",
        "print(\"Recall Score %0.2f\" %(recall_score(y_test,y_predicted)*100))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Accuracy Test 81.61\n",
            "Accuracy Train 93.20\n",
            "ROC-AUC Score 89.81\n",
            "Recall Score 76.47\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "id": "PE-_G7Wpzwmu",
        "outputId": "773cf000-a6d6-493a-9696-f04d896dc1a3"
      },
      "source": [
        "plt.bar(range(len(clf.feature_importances_)),clf.feature_importances_)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<BarContainer object of 25 artists>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 15
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "nPqoZNpH2Iqp",
        "outputId": "b040f43b-b1cf-42a5-fe49-20675f4ba176"
      },
      "source": [
        "from xgboost import XGBRFClassifier\n",
        "xgb = XGBRFClassifier()\n",
        "xgb.fit(x_train,y_train)\n",
        "y_predicted = xgb.predict(x_test)\n",
        "y_train_pred = xgb.predict(x_train)\n",
        "\n",
        "pred_prob = xgb.predict_proba(x_test)\n",
        "print(\"Accuracy Test %0.2f\" %(accuracy_score(y_test,y_predicted)*100))\n",
        "print(\"Accuracy Train %0.2f\" %(accuracy_score(y_train,y_train_pred)*100))\n",
        "print(\"ROC-AUC Score %0.2f\" %(roc_auc_score(y_test,pred_prob[:,1])*100))\n",
        "print(\"Recall Score %0.2f\" %(recall_score(y_test,y_predicted)*100))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Accuracy Test 77.01\n",
            "Accuracy Train 82.60\n",
            "ROC-AUC Score 86.98\n",
            "Recall Score 76.47\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 381
        },
        "id": "VK7PO8Zd2Ri2",
        "outputId": "a7694b6a-ffbf-4b8c-c75f-1d8bf4a8b335"
      },
      "source": [
        "plt.bar(range(len(xgb.feature_importances_)),xgb.feature_importances_)\n",
        "plt.title(\"Feature Importance for XGBoost\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Feature Importance for XGBoost')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeEAAAFZCAYAAACv05cWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAdSUlEQVR4nO3de1DVdf7H8deB4yUDkYMcc7WSLNFordVdL2GSJKaN1mZtoAl2s8w1I9efEmpoCrnm1ka5U5o2Wc2KF7atscLKckrx0mU0cTbTLRZL5SqKGgp8fn84npEirgc+cno+ZnbiXL/v76dv+/R8z+HoMMYYAQCAFudnewAAAH6tiDAAAJYQYQAALCHCAABYQoQBALCECAMAYAkRxgUnPDxcMTExGjlypOd/999/f6Of7/Tp03rzzTe9OOHPhYeH6/Dhw826jZoUFhbqww8/bPHtnm/Xrl2KiorS5MmTG/0cX3/9tQYOHKjvv//ec11BQYEGDRqkPXv2SJKMMVq1apVuvfVWjRo1SjExMZo0aZLndklKSkrSoEGDPMfNmDFj9Pbbbzd+52pxIaw9Wj+n7QGAmrz22mu65JJLvPJce/fu1Ztvvqk//vGPXnm+C8n27du1detW3XTTTdZm+PTTTzVgwAA9/fTTjX6O8PBwTZw4UU888YRWrFghSVqwYIFiY2N1zTXXSJKeffZZbd++XS+//LLcbrcqKyu1du1a3XvvvcrKypLL5ZIkJSQkaMqUKZKkffv26c4779TgwYPVuXPnJu5pdRfC2qP1I8JoVQ4fPqx58+bp22+/lSQlJycrKipKkrR27VqtXLlSlZWVCg0N1eLFi9WuXTtNnTpVZWVlGj9+vBYvXqwRI0Zo7969kqSDBw96LmdmZmrTpk06fvy4IiIiNHPmTGVkZOiVV17R6dOndd111yktLU3t27evdcbo6Gjde++9yszM1JEjRzRv3jxlZ2frk08+kcvl0vLlyxUUFKTw8HDNnj1b69evV35+vqZNm6Zx48ZJklatWqXVq1erqqpKYWFhSk1NlcvlUlJSkoKCgrR161aNGTNGK1asUGVlpU6ePKlnn322xjXo1q2bMjMz9fHHHysgIECff/65/P399dxzz+mqq65ScXGxkpOT9c0336hDhw6aNWuWhgwZomPHjmnBggXavXu3KioqNGXKFN1xxx3V9vW9997TqlWrVFlZqUmTJmn58uX1mn3KlCkaNWpUted68MEHdddddykzM1PBwcE6cOCAlixZIkk6evSoXn31Vf373/+W2+2WJPn7+ysuLk6jR49WQEBAjf8uevXqpaCgIP3www/q3LmzfvjhB82dO1cHDx5UmzZt9MADD3j+cPbuu+9q6dKlqqiokNvt1sKFC3XZZZdp3759mjt3rsrKynTmzBklJCTod7/7nZ588slqaw80igEuML169TKHDh2q8baEhATz7LPPGmOM+e6778yAAQNMcXGxKSwsNNdcc43ncUlJSSY5OdkYY8z69evNxIkTjTHG5OXlmT59+nie7/zL69evN9ddd5359ttvjTHG7Ny50wwePNgcPnzYGGPM3LlzzaJFi+qcediwYWbu3LnGGGNee+01c+2115pt27aZqqoqc8cdd5g1a9Z4HvPkk08aY4w5cOCAueaaa0xxcbH58ssvzdChQ01hYaExxpgnn3zSsy+zZs0yY8aMMT/++KMxxpj09HTPbXWtwbXXXmu++uorY4wx8+bNM7NnzzbGGJOcnGwWL15sjDEmJyfHDBgwwJSXl5vHH3/czJw501RWVpqioiITFRVlvv7665/t+/kzNGT2mvznP/8xkZGRZtiwYWbXrl2e6z/++GNz8803/+Ljzpk1a5ZZunRptcdFR0eb8vJyY4wx9913n3nxxReNMcYcPHjQ9O/f3+Tl5Znvv//e9O/f33z33XfGGGNWrFjhOWYeeeQRk5mZaYwxpqioyDz88MOmvLy82n4DjcV7wrggxcfHV3tPeM6cOTp58qS2b9+ue+65R5J0+eWXq3///tq8ebNCQkL0+eefe05h//73v1deXl6Dt9ujRw/16NFDkrRp0ybdcsst6tKliyRp3Lhx2rhxY72e59wpyl69eqldu3YaOHCgHA6HrrrqKuXn53vud+6V5RVXXKGwsDDt3r1bH3/8sW6++WaFhIRIkv70pz9py5YtnscMHjxY7dq1+9k261qDnj17ek7tXn311Tp06JAkafPmzRo9erTn+g8//FBt27bVRx99pISEBPn5+cnlcikmJqbO/W/s7OeEh4erW7du8vf319VXX+25vrS01HO6WZKOHTvmOTaGDh2q5cuXe25btWqVRo4cqRtvvFHTpk3TQw89pLZt2+rMmTPaunWrxo8fL0nq1q2bBg4cqG3btmnLli0aOHCgLr/8cs/c27dvV0VFhUJCQpSVlaWcnBwFBwfrH//4h9q2bVvrOgD1xeloXJBqek/4yJEjMsYoLi7Oc93Jkyc1aNAgVVZWKj09XZs2bVJlZaVOnDihsLCwBm83KCjI8/Px48f1/vvv69NPP5V09oNBZ86cqdfzXHzxxZIkPz8/z8/nLldVVdW4vaCgIB07dkzFxcWeU66S1LFjRxUVFdX4mPPVtQaBgYGen/39/VVZWSnp7Kne8287d2r3+PHjSkxMlL+/vySpvLxcI0eOrHW/Gzv7OevWrVO7du105ZVX6uWXX/Z82MvlclX7w0vHjh313nvvSZJmz56tH3/80XPb+e8JFxcXKzExUWfOnNGIESNkjKm2rx07dlRxcbHn53MCAwNljFFJSYlmzJihl156SYmJiSovL9dDDz2ku+++u9b9AOqLCKPVCAkJkb+/v9avX18tbJL09ttva9OmTXr99dflcrm0Zs2aGj8V6+/vr6qqKhlj5HA4dOzYsV/cntvt1u23365Zs2Z5fV/OKSkpUbdu3SSdjWFQUJA6d+6so0ePeu5z9OjRen2o6J133qnXGvxUp06dVFJSou7du0s6+z55ly5d5Ha7tXTpUvXq1ave+9PY2aWz7/f//e9/1xtvvKG2bdvq9ttv14gRI3TFFVfouuuuU1FRkfbu3VvtFXJdXC6XRo0apY8++kixsbHy8/NTaWmp5w8DR48e9bxq//LLLz2PKy0tlZ+fn4KDg+V0OjV9+nRNnz5du3fv1qRJk3T99dfXewagNpyORqvhdDoVFRWl1atXS5JOnTqlxx9/XIcOHVJRUZG6desml8ulkpISvfvuuzpx4oTncWVlZTLGKDg4WP7+/vr6668lqdZfXYqOjtbGjRs9r5Q++OADLVu2zKv7tGHDBknSgQMHlJubq2uvvVY33nij3n//fZWUlEiSVq9e7fnw2U85nU4dP35ckmpdg9pER0frX//6lyRp//79Gjt2rCorKxUdHe1Z64qKCqWlpSknJ6fW52rI7D81e/ZsTZw4UZdffrm6du2qhx9+WHPmzJExRgEBAZoyZYpmzpyp3NxcSVJVVZU2bNigd999V5dddlmNz3n69Glt3rxZV155pZxOp4YMGaKMjAxJ0v/+9z999tlnuv766xUZGanPPvvMc/p+9erVioyMlNPp1OTJk/XNN99IOvv2QkBAgBwOR7W1BxqLV8JoVebNm6eUlBStXbtWknTrrbeqa9euGj16tDZs2KCYmBhdeumlSkxM1MMPP6xFixYpPj5eS5Ys0Q033KDNmzfrkUce0QMPPCC32634+Phf3FZERIQmT56s+Ph4VVVVKSQkRPPnz/fq/rhcLt122206cuSI5syZo6CgIPXt21cPPvig7r77blVVValPnz6aN29ejY+PjIzUK6+8ojvuuEMvvfTSL65Bba9m/+///k+zZs1SdHS0Lr74Yi1ZskTt27dXYmKi5s+fr5tvvlmSdMMNNyg8PLzW/WnI7OfLyMhQcXGx7r33Xs918fHxevvtt/XGG29owoQJmjRpkjp16qRp06apvLxcp0+fVlhYmNLT0zVkyBDP41atWqW33npL0tlT9EOGDFFiYqIkaf78+ZozZ44yMzPVpk0bLVy4UF27dpUkLVy4UFOmTNGZM2fUvXt3LViwQJI0YcIE/eUvf/G8FTF+/Hj16NGj2tqvX7++zn0EauIwhr9PGLAhPDxcmzdv9trvQwNofTgdDQCAJUQYAABLOB0NAIAlvBIGAMASIgwAgCUt/itKBQUt93t1wcEdVFJyssW256tYR+9gHZuONfQO1tE7GrKOoaGBNV7v06+EnU5/2yP4BNbRO1jHpmMNvYN19A5vrKNPRxgAgAsZEQYAwBIiDACAJUQYAABLiDAAAJYQYQAALCHCAABYQoQBALCECAMAYAkRBgDAEiIMAIAlRBgAAEta/G9RAgCgvu5btKnBj1mZFN0MkzQPXgkDAGAJEQYAwBIiDACAJUQYAABLiDAAAJYQYQAALCHCAABYQoQBALCECAMAYAkRBgDAEiIMAIAlRBgAAEuIMAAAlhBhAAAsIcIAAFhChAEAsIQIAwBgCREGAMASIgwAgCVEGAAAS4gwAACWEGEAACwhwgAAWEKEAQCwhAgDAGAJEQYAwBIiDACAJUQYAABLiDAAAJYQYQAALCHCAABYQoQBALCECAMAYEm9IpyWlqbY2FjFxcVp9+7dNd7nb3/7m+Lj4706HAAAvqzOCO/YsUO5ubnKyMhQamqqUlNTf3af/fv3a+fOnc0yIAAAvqrOCGdnZ2v48OGSpJ49e6q0tFRlZWXV7rNo0SI99thjzTMhAAA+ylnXHQoLCxUREeG57HK5VFBQoICAAElSZmamBgwYoG7dutVrg8HBHeR0+jdy3IYLDQ1ssW35MtbRO1jHpmMNvcOX17El962p26ozwj9ljPH8fPToUWVmZuqVV17RkSNH6vX4kpKTDd1ko4WGBqqg4HiLbc9XsY7ewTo2HWvoHb6+ji21bw1Zx1+KdZ2no91utwoLCz2X8/PzFRoaKknatm2biouLdffdd2vq1KnKyclRWlpavQYCAODXrs4IR0ZGKisrS5KUk5Mjt9vtORU9cuRIvfPOO1qzZo1eeOEFRUREKDk5uXknBgDAR9R5Orpfv36KiIhQXFycHA6HUlJSlJmZqcDAQMXExLTEjAAA+KR6vSc8Y8aMapd79+79s/t0795dr732mnemAgDgV4BvzAIAwBIiDACAJUQYAABLiDAAAJYQYQAALCHCAABYQoQBALCECAMAYAkRBgDAEiIMAIAlRBgAAEuIMAAAlhBhAAAsIcIAAFhChAEAsIQIAwBgCREGAMASIgwAgCVEGAAAS4gwAACWEGEAACwhwgAAWEKEAQCwhAgDAGAJEQYAwBIiDACAJUQYAABLiDAAAJYQYQAALCHCAABYQoQBALCECAMAYAkRBgDAEiIMAIAlRBgAAEuIMAAAlhBhAAAsIcIAAFhChAEAsIQIAwBgCREGAMASIgwAgCVEGAAAS4gwAACWEGEAACwhwgAAWEKEAQCwhAgDAGAJEQYAwBIiDACAJUQYAABLiDAAAJYQYQAALHHW505paWnatWuXHA6HkpOT1bdvX89ta9as0bp16+Tn56fevXsrJSVFDoej2QYGAMBX1PlKeMeOHcrNzVVGRoZSU1OVmprque3UqVPasGGD3njjDa1evVr//e9/9eWXXzbrwAAA+Io6I5ydna3hw4dLknr27KnS0lKVlZVJki666CK9+uqratOmjU6dOqWysjKFhoY278QAAPiIOk9HFxYWKiIiwnPZ5XKpoKBAAQEBnuuWLVumVatWKSEhQZdeemmtzxcc3EFOp38TRm6Y0NDAFtuWL2MdvYN1bDrW0Dt8eR1bct+auq16vSd8PmPMz6578MEHlZCQoEmTJql///7q37//Lz6+pORkQzfZaKGhgSooON5i2/NVrKN3sI5Nxxp6h6+vY0vtW0PW8ZdiXefpaLfbrcLCQs/l/Px8zynno0ePaufOnZKk9u3ba+jQofriiy/qNRAAAL92dUY4MjJSWVlZkqScnBy53W7PqeiKigolJSXpxIkTkqSvvvpKYWFhzTguAAC+o87T0f369VNERITi4uLkcDiUkpKizMxMBQYGKiYmRn/+85+VkJAgp9Op8PBw3XTTTS0xNwAArV693hOeMWNGtcu9e/f2/Dx27FiNHTvWu1MBAPArwDdmAQBgCREGAMASIgwAgCVEGAAAS4gwAACWEGEAACwhwgAAWEKEAQCwhAgDAGAJEQYAwBIiDACAJUQYAABLiDAAAJYQYQAALCHCAABYQoQBALCECAMAYAkRBgDAEiIMAIAlRBgAAEuIMAAAlhBhAAAsIcIAAFhChAEAsIQIAwBgCREGAMASIgwAgCVEGAAAS4gwAACWEGEAACwhwgAAWEKEAQCwhAgDAGAJEQYAwBIiDACAJUQYAABLiDAAAJYQYQAALCHCAABYQoQBALCECAMAYAkRBgDAEiIMAIAlRBgAAEuIMAAAlhBhAAAsIcIAAFhChAEAsIQIAwBgCREGAMASIgwAgCVEGAAAS4gwAACWOOtzp7S0NO3atUsOh0PJycnq27ev57Zt27bpmWeekZ+fn8LCwpSamio/P9oOAEBd6qzljh07lJubq4yMDKWmpio1NbXa7U888YTS09O1evVqnThxQp988kmzDQsAgC+pM8LZ2dkaPny4JKlnz54qLS1VWVmZ5/bMzExdcsklkiSXy6WSkpJmGhUAAN9S5+nowsJCRUREeC67XC4VFBQoICBAkjz/zM/P15YtW/Too4/W+nzBwR3kdPo3ZeYGCQ0NbLFt+TLW0TtYx6ZjDb3Dl9exJfetqduq13vC5zPG/Oy6oqIiTZ48WSkpKQoODq718SUlJxu6yUYLDQ1UQcHxFtuer2IdvYN1bDrW0Dt8fR1bat8aso6/FOs6T0e73W4VFhZ6Lufn5ys0NNRzuaysTJMmTVJiYqKGDBlSr2EAAEA9IhwZGamsrCxJUk5Ojtxut+cUtCQtWrRIEydO1NChQ5tvSgAAfFCdp6P79euniIgIxcXFyeFwKCUlRZmZmQoMDNSQIUP05ptvKjc3V+vWrZMkjR49WrGxsc0+OAAArV293hOeMWNGtcu9e/f2/Lxnzx7vTgQAwK8E36oBAIAlRBgAAEuIMAAAlhBhAAAsIcIAAFhChAEAsIQIAwBgCREGAMASIgwAgCVEGAAAS4gwAACWEGEAACwhwgAAWEKEAQCwhAgDAGAJEQYAwBIiDACAJUQYAABLnLYHAACgudy3aFODH7MyKboZJqkZr4QBALCECAMAYAkRBgDAEiIMAIAlRBgAAEuIMAAAlhBhAAAsIcIAAFhChAEAsIQIAwBgCREGAMASIgwAgCVEGAAAS4gwAACWEGEAACwhwgAAWEKEAQCwhAgDAGAJEQYAwBIiDACAJUQYAABLiDAAAJYQYQAALCHCAABYQoQBALCECAMAYAkRBgDAEiIMAIAlRBgAAEuIMAAAlhBhAAAsIcIAAFhChAEAsKReEU5LS1NsbKzi4uK0e/fuareVl5dr1qxZGjt2bLMMCACAr6ozwjt27FBubq4yMjKUmpqq1NTUarcvXrxYffr0abYBAQDwVXVGODs7W8OHD5ck9ezZU6WlpSorK/Pc/thjj3luBwAA9VdnhAsLCxUcHOy57HK5VFBQ4LkcEBDQPJMBAODjnA19gDGmSRsMDu4gp9O/Sc/REKGhgS22LV/GOnoH69h0rKF3+PI6NnXfGvL4pm6rzgi73W4VFhZ6Lufn5ys0NLTRGywpOdnoxzZUaGigCgqOt9j2fBXr6B2sY9Oxht7h6+vY1H2r7+Mbso6/FOs6T0dHRkYqKytLkpSTkyO3280paAAAvKDOV8L9+vVTRESE4uLi5HA4lJKSoszMTAUGBiomJkbTpk3T4cOH9e233yo+Pl533XWXxowZ0xKzAwDQqtXrPeEZM2ZUu9y7d2/Pz+np6d6dCACAXwm+MQsAAEuIMAAAlhBhAAAsIcIAAFhChAEAsIQIAwBgCREGAMASIgwAgCVEGAAAS4gwAACWEGEAACwhwgAAWFKvv8ABtbtv0aYGP2ZlUnQzTAIAaE14JQwAgCVEGAAAS4gwAACWEGEAACwhwgAAWEKEAQCwhAgDAGAJEQYAwBIiDACAJUQYAABLiDAAAJYQYQAALCHCAABYQoQBALCECAMAYAkRBgDAEiIMAIAlRBgAAEuctgfAWfct2tTgx6xMim6GSQAALYUIw2v4gwQANAynowEAsIRXwoAXcTYAQEMQYfF/nIC38d8UUD+cjgYAwBIiDACAJUQYAABLiDAAAJa0+g9mNeYDIBIfAgFwYbsQPtx2Iczg63glDACAJUQYAABLiDAAAJYQYQAALCHCAABY0uo/HY2zmvopRj4FCQAtj1fCAABYQoQBALCECAMAYAkRBgDAEiIMAIAlRBgAAEvqFeG0tDTFxsYqLi5Ou3fvrnbb1q1bdeeddyo2NlZLly5tliEBAPBFdUZ4x44dys3NVUZGhlJTU5Wamlrt9oULF+r555/XP//5T23ZskX79+9vtmEBAPAldUY4Oztbw4cPlyT17NlTpaWlKisrkyTl5eUpKChIXbt2lZ+fn6KiopSdnd28EwMA4CPqjHBhYaGCg4M9l10ulwoKCiRJBQUFcrlcNd4GAABq5zDGmNruMHfuXEVFRXleDY8bN05paWkKCwvTF198oRUrVnjeC167dq3y8vI0ffr05p8cAIBWrs5Xwm63W4WFhZ7L+fn5Cg0NrfG2I0eOyO12N8OYAAD4njojHBkZqaysLElSTk6O3G63AgICJEndu3dXWVmZDh48qIqKCn300UeKjIxs3okBAPARdZ6OlqQlS5bos88+k8PhUEpKivbu3avAwEDFxMRo586dWrJkiSRpxIgRuv/++5t9aAAAfEG9IgwAALyPb8wCAMASIgwAgCVO2wM0l7S0NO3atUsOh0PJycnq27ev7ZFale3bt+vRRx/VVVddJUnq1auX5s6da3mq1mXfvn2aMmWK7rnnHk2YMEGHDh3SzJkzVVlZqdDQUD399NNq27at7TEveD9dx6SkJOXk5KhTp06SpPvvv1833nij3SEvcIsXL9bnn3+uiooKPfTQQ/rtb3/LsdgIP13HTZs2NflY9MkIn/9VmwcOHFBycrIyMjJsj9XqDBgwQOnp6bbHaJVOnjypBQsWaPDgwZ7r0tPTNX78eI0aNUrPPPOM1q1bp/Hjx1uc8sJX0zpK0vTp0zVs2DBLU7Uu27Zt0zfffKOMjAyVlJTo9ttv1+DBgzkWG6imdRw0aFCTj0WfPB1d21dtAi2hbdu2Wr58ebXfm9++fbtuuukmSdKwYcP4itd6qGkd0TB/+MMf9Nxzz0mSOnbsqFOnTnEsNkJN61hZWdnk5/XJCNf2VZuov/3792vy5MkaN26ctmzZYnucVsXpdKp9+/bVrjt16pTnlF9ISAjHZD3UtI6S9PrrryshIUGPPfaYiouLLUzWevj7+6tDhw6SpHXr1mno0KEci41Q0zr6+/s3+Vj0ydPRP8VvYTVcjx49NHXqVI0aNUp5eXlKSEjQxo0bed/ISzgmG++2225Tp06d1KdPHy1btkwvvPCCnnjiCdtjXfA++OADrVu3TitXrtSIESM813MsNsz567hnz54mH4s++Uq4tq/aRP106dJFt9xyixwOhy677DJ17txZR44csT1Wq9ahQwf9+OOPkviK16YYPHiw+vTpI0mKjo7Wvn37LE904fvkk0/04osvavny5QoMDORYbKSfrqM3jkWfjHBtX7WJ+nnrrbe0YsUKSWf/tqyioiJ16dLF8lSt2/XXX+85Ljdu3KgbbrjB8kSt0yOPPKK8vDxJZ99nP/cJftTs+PHjWrx4sV566SXPp3g5FhuupnX0xrHos9+Y9dOv2uzdu7ftkVqVsrIyzZgxQ8eOHdOZM2c0depURUVF2R6r1dizZ4/++te/6vvvv5fT6VSXLl20ZMkSJSUlqby8XL/5zW/01FNPqU2bNrZHvaDVtI4TJkzQsmXLdNFFF6lDhw566qmnFBISYnvUC1ZGRoaef/55hYWFea5btGiR5syZw7HYADWt49ixY/X666836Vj02QgDAHCh88nT0QAAtAZEGAAAS4gwAACWEGEAACwhwgAAWEKEAQCwhAgDAGAJEQYAwJL/B/xaQTffbqfQAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "14WYKXed0OlN",
        "outputId": "37fb1650-5e73-4ca0-accd-57882b691c0a"
      },
      "source": [
        "from lightgbm import LGBMClassifier\n",
        "lgb = LGBMClassifier()\n",
        "lgb.fit(x_train,y_train)\n",
        "y_predicted = lgb.predict(x_test)\n",
        "y_train_pred = lgb.predict(x_train)\n",
        "\n",
        "pred_prob = lgb.predict_proba(x_test)\n",
        "print(\"Accuracy Test %0.2f\" %(accuracy_score(y_test,y_predicted)*100))\n",
        "print(\"Accuracy Train %0.2f\" %(accuracy_score(y_train,y_train_pred)*100))\n",
        "print(\"ROC-AUC Score %0.2f\" %(roc_auc_score(y_test,pred_prob[:,1])*100))\n",
        "print(\"Recall Score %0.2f\" %(recall_score(y_test,y_predicted)*100))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Accuracy Test 77.01\n",
            "Accuracy Train 100.00\n",
            "ROC-AUC Score 85.51\n",
            "Recall Score 74.51\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 381
        },
        "id": "hQLizVa-1atq",
        "outputId": "4d788ca5-171d-41af-c452-b9164a9317ca"
      },
      "source": [
        "plt.bar(range(len(lgb.feature_importances_)),lgb.feature_importances_)\n",
        "plt.title(\"Feature Importance for LightGBM\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Feature Importance for LightGBM')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 20
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VnB4Y4iD1FSd"
      },
      "source": [
        "!pip install catboost"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oo-hQO9q04GG"
      },
      "source": [
        "from catboost import CatBoostClassifier\n",
        "cgb = CatBoostClassifier(iterations=1000)\n",
        "cgb.fit(x_train,y_train)\n",
        "y_predicted = cgb.predict(x_test)\n",
        "y_train_pred = cgb.predict(x_train)\n",
        "pred_prob = cgb.predict_proba(x_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HgTAnl6U1UmD",
        "outputId": "cf86ed8b-6a97-47b4-c5d6-939ad56250fe"
      },
      "source": [
        "print(\"Accuracy Test %0.2f\" %(accuracy_score(y_test,y_predicted)*100))\n",
        "print(\"Accuracy Train %0.2f\" %(accuracy_score(y_train,y_train_pred)*100))\n",
        "print(\"ROC-AUC Score %0.2f\" %(roc_auc_score(y_test,pred_prob[:,1])*100))\n",
        "print(\"Recall Score %0.2f\" %(recall_score(y_test,y_predicted)*100))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Accuracy Test 80.46\n",
            "Accuracy Train 95.60\n",
            "ROC-AUC Score 89.76\n",
            "Recall Score 76.47\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 381
        },
        "id": "bJLkPu9D1n8V",
        "outputId": "b2a686dd-dc6d-49ee-c7c9-5475e8910d2e"
      },
      "source": [
        "plt.bar(range(len(cgb.feature_importances_)),cgb.feature_importances_)\n",
        "plt.title(\"Feature Importance for CatBoost\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Feature Importance for CatBoost')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 83
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "WEYDO86Z32tZ"
      },
      "source": [
        "Least Important Features : 9,10,12,13,15,16,17,18,20,22,25\n",
        "\n",
        "Modeling by removal of these features."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JOlhLyZD6frP"
      },
      "source": [
        "df = pd.DataFrame(df.to_numpy())"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FAfdrj_yALED"
      },
      "source": [
        "df.drop(columns=[9,10,12,13,15,16,17,18,20,22,25],inplace=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KJmc5YnH4q1L"
      },
      "source": [
        "X = df.to_numpy()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "SyZt1CMBAXao",
        "outputId": "882e52f8-edbe-48d5-d3fe-9362008c92d7"
      },
      "source": [
        "X.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(587, 15)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GqA6YH0FBMdf"
      },
      "source": [
        "Y = Y.astype(int)\n",
        "Y = Y.reshape(-1)\n",
        "x_train, x_test, y_train, y_test = train_test_split(X,Y,test_size = 0.148)\n",
        "scaler = preprocessing.StandardScaler().fit(x_train)\n",
        "x_train = scaler.transform(x_train)\n",
        "scaler = preprocessing.StandardScaler().fit(x_test)\n",
        "x_test = scaler.transform(x_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gwcTdN-2AjNK"
      },
      "source": [
        "least_imp = [9,10,12,13,15,16,17,18,20,22]\n",
        "x_train = np.delete(x_train,least_imp,axis=1)\n",
        "x_test = np.delete(x_test,least_imp,axis=1)\n",
        "clf = RandomForestClassifier(n_estimators=1333,bootstrap=True,max_depth=80, max_features='sqrt', min_samples_split= 4, min_samples_leaf=3).fit(x_train,y_train)\n",
        "clf.fit(x_train,y_train)\n",
        "y_predicted = clf.predict(x_test)\n",
        "y_train_pred = clf.predict(x_train)\n",
        "\n",
        "pred_prob = clf.predict_proba(x_test)\n",
        "print(\"Accuracy Test %0.2f\" %(accuracy_score(y_test,y_predicted)*100))\n",
        "print(\"Accuracy Train %0.2f\" %(accuracy_score(y_train,y_train_pred)*100))\n",
        "print(\"ROC-AUC Score %0.2f\" %(roc_auc_score(y_test,pred_prob[:,1])*100))\n",
        "print(\"Recall Score %0.2f\" %(recall_score(y_test,y_predicted)*100))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "id": "wIIlEizvAtyj",
        "outputId": "544af1ce-a6b4-4396-aaab-1c10d854f43e"
      },
      "source": [
        "plt.bar(range(len(clf.feature_importances_)),clf.feature_importances_)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<BarContainer object of 9 artists>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 32
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PdHqisLgCCC9",
        "outputId": "f6f50fc5-19ac-469f-ee6e-d8295216618e"
      },
      "source": [
        "from xgboost import XGBRFClassifier\n",
        "xgb = XGBRFClassifier()\n",
        "xgb.fit(x_train,y_train)\n",
        "y_predicted = xgb.predict(x_test)\n",
        "y_train_pred = xgb.predict(x_train)\n",
        "\n",
        "pred_prob = xgb.predict_proba(x_test)\n",
        "print(\"Accuracy Test %0.2f\" %(accuracy_score(y_test,y_predicted)*100))\n",
        "print(\"Accuracy Train %0.2f\" %(accuracy_score(y_train,y_train_pred)*100))\n",
        "print(\"ROC-AUC Score %0.2f\" %(roc_auc_score(y_test,pred_prob[:,1])*100))\n",
        "print(\"Recall Score %0.2f\" %(recall_score(y_test,y_predicted)*100))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Accuracy Test 85.06\n",
            "Accuracy Train 83.60\n",
            "ROC-AUC Score 88.57\n",
            "Recall Score 95.65\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 381
        },
        "id": "iP5_QrdmCKJ8",
        "outputId": "c3d7a2cd-9e55-415e-b0c6-d51301cb0fc7"
      },
      "source": [
        "plt.bar(range(len(xgb.feature_importances_)),xgb.feature_importances_)\n",
        "plt.title(\"Feature Importance for XGBoost\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Feature Importance for XGBoost')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 148
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gxx4KrA0CbAB",
        "outputId": "8bc8013a-de26-47bc-dccb-a8ee35af2cce"
      },
      "source": [
        "from lightgbm import LGBMClassifier\n",
        "lgb = LGBMClassifier()\n",
        "lgb.fit(x_train,y_train)\n",
        "y_predicted = lgb.predict(x_test)\n",
        "y_train_pred = lgb.predict(x_train)\n",
        "\n",
        "pred_prob = lgb.predict_proba(x_test)\n",
        "print(\"Accuracy Test %0.2f\" %(accuracy_score(y_test,y_predicted)*100))\n",
        "print(\"Accuracy Train %0.2f\" %(accuracy_score(y_train,y_train_pred)*100))\n",
        "print(\"ROC-AUC Score %0.2f\" %(roc_auc_score(y_test,pred_prob[:,1])*100))\n",
        "print(\"Recall Score %0.2f\" %(recall_score(y_test,y_predicted)*100))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Accuracy Test 82.76\n",
            "Accuracy Train 100.00\n",
            "ROC-AUC Score 88.49\n",
            "Recall Score 89.86\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 381
        },
        "id": "LWXYEJITCdAt",
        "outputId": "7dc22485-ec2a-4245-dcb9-e3285daea442"
      },
      "source": [
        "plt.bar(range(len(df.columns)),lgb.feature_importances_)\n",
        "plt.title(\"Feature Importance for LightGBM\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Feature Importance for LightGBM')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 150
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FmVUwcqoI3cW",
        "outputId": "2a596096-f1ac-4764-8b33-a945ac3a9a64"
      },
      "source": [
        "# Removing all the least important features.\n",
        "least_imp = [9,10,12,13,15,16,17,18,20,22]\n",
        "for each in least_imp:\n",
        "    del features[each]\n",
        "features"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{0: 'pulse',\n",
              " 1: 'respiration',\n",
              " 2: 'systolic.blood.pressure',\n",
              " 3: 'diastolic.blood.pressure',\n",
              " 4: 'map',\n",
              " 5: 'BMI',\n",
              " 6: 'NYHA.cardiac.function.classification',\n",
              " 7: 'Killip.grade',\n",
              " 8: 'myocardial.infarction',\n",
              " 11: 'Chronic.obstructive.pulmonary.disease',\n",
              " 14: 'moderate.to.severe.chronic.kidney.disease',\n",
              " 19: 'CCI.score',\n",
              " 21: 'fio2',\n",
              " 23: 'left.ventricular.end.diastolic.diameter.LV',\n",
              " 24: 'brain.natriuretic.peptide'}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 161
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QFrWpcTjMmYX"
      },
      "source": [
        "Successful Selection of important features\n",
        "```\n",
        "With 25 Features on RFClassifier\n",
        "Accuracy Test  87.36\n",
        "Accuracy Train 94.20\n",
        "ROC-AUC Score  88.81\n",
        "Recall Score   92.75\n",
        "\n",
        "With 15 Features (Most imp) on RFClassifier\n",
        "Accuracy Test 87.36\n",
        "Accuracy Train 95.00\n",
        "ROC-AUC Score 88.81\n",
        "Recall Score 92.75\n",
        "```\n",
        "Validated by training on 25 features and then removing the leat_imp features from the same set and training again."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VPEO64GpPnWy"
      },
      "source": [
        "frame = pd.DataFrame(X)\n",
        "#frame.drop(columns=[9,10,12,13,15,16,17,18,20,22])\n",
        "frame['lvef_class'] = Y\n",
        "frame.to_csv(\"Reduced_Features_set.csv\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cdMfo9PKcPjl",
        "outputId": "84cc73ea-9466-47e5-da3c-2c41e209769c"
      },
      "source": [
        "frame.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(587, 16)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "fBcb1DNPXx_K"
      },
      "source": [
        "Most Important Features :\n",
        "1. Brain Natriuretic Peptide\n",
        "2. LV end diastolic diameter"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "icdlNxBPbrkk"
      },
      "source": [
        "----"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YwANAXCrbwqE"
      },
      "source": [
        "Testing the Stacking classifier over reduced featues"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HzCquQX5hVHl",
        "outputId": "2cd13c02-cfb8-4345-9f96-4ee1084d6b46"
      },
      "source": [
        "x_train.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(500, 15)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 195
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OAHCm_y2bwRB"
      },
      "source": [
        "from sklearn.ensemble import StackingClassifier\n",
        "clf = RandomForestClassifier(n_estimators=1333,bootstrap=True,max_depth=80, max_features='sqrt', min_samples_split= 4, min_samples_leaf=3)\n",
        "estimators = [('rfc',clf),('lgb',LGBMClassifier()),('cat',CatBoostClassifier(iterations=5000),('xgb',XGBRFClassifier()))]\n",
        "final_estimator = RandomForestClassifier()\n",
        "\n",
        "clf_stack = StackingClassifier(estimators, final_estimator)\n",
        "clf_stack.fit(x_train,y_train)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "c1ICcJsqhDPL",
        "outputId": "4ecb6e85-c214-4d5a-c53a-3edfe81c05ee"
      },
      "source": [
        "y_predicted = clf_stack.predict(x_test)\n",
        "y_train_pred = clf_stack.predict(x_train)\n",
        "pred_prob = clf_stack.predict_proba(x_test)\n",
        "print(\"Accuracy Test %0.2f\" %(accuracy_score(y_test,y_predicted)*100))\n",
        "print(\"Accuracy Train %0.2f\" %(accuracy_score(y_train,y_train_pred)*100))\n",
        "print(\"ROC-AUC Score %0.2f\" %(roc_auc_score(y_test,pred_prob[:,1])*100))\n",
        "print(\"Recall Score %0.2f\" %(recall_score(y_test,y_predicted)*100))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Accuracy Test 83.91\n",
            "Accuracy Train 91.80\n",
            "ROC-AUC Score 87.72\n",
            "Recall Score 94.20\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oDKf2XAskwoI",
        "outputId": "28aac709-f762-4e6b-cffc-786a856709d1"
      },
      "source": [
        "len(np.where(Y==0)[0])\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "140"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 232
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5OqStySvmS-M"
      },
      "source": [
        "SOLVE THIS ISSUE OF IMBALANCE"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 297
        },
        "id": "KieLSgE7jUST",
        "outputId": "63aa9529-c1f3-4058-f925-c892a5ec902f"
      },
      "source": [
        "fig, axes = plt.subplots(nrows=1, ncols=2, figsize=(15, 5))\n",
        "axes[0].scatter(x_test[:,4],x_test[:,13],c=y_predicted,cmap='brg',s=25)\n",
        "axes[1].scatter(x_test[:,4],x_test[:,13],c=y_test,cmap='brg',s=25)\n",
        "axes[0].title.set_text(\"Predictions\")\n",
        "axes[1].title.set_text(\"Actual Plot\")\n",
        "fig.tight_layout()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1080x360 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cEiLRKd8gAsC"
      },
      "source": [
        "Included new Features for explorations :\n",
        "* High Senstiivity Troponin\n",
        "* Cholestrol\n",
        "* Creatine\n",
        "* Cystanin"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Y7W6zSsPgAbF"
      },
      "source": [
        "df = pd.read_csv(\"/content/07_with_LVEF_sodium_HST_BNP.csv\")\n",
        "df.drop(columns=['Unnamed: 0','map','diabetes','type.of.heart.failure','liver.disease','congestive.heart.failure','acute.renal.failure','moderate.to.severe.chronic.kidney.disease','dementia','peptic.ulcer.disease','hemiplegia','connective.tissue.disease','leukemia','AIDS','weight','height'],inplace=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gLICTN4aliq4",
        "outputId": "f30c6ca8-b283-4e7d-b709-4a27b2ee07f7"
      },
      "source": [
        "df.columns"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['pulse', 'respiration', 'systolic.blood.pressure',\n",
              "       'diastolic.blood.pressure', 'map', 'BMI',\n",
              "       'NYHA.cardiac.function.classification', 'Killip.grade',\n",
              "       'myocardial.infarction', 'peripheral.vascular.disease',\n",
              "       'cerebrovascular.disease', 'Chronic.obstructive.pulmonary.disease',\n",
              "       'diabetes', 'CCI.score', 'type.II.respiratory.failure', 'fio2', 'LVEF',\n",
              "       'left.ventricular.end.diastolic.diameter.LV',\n",
              "       'brain.natriuretic.peptide', 'cystatin', 'creatine.kinase',\n",
              "       'cholesterol', 'creatinine.enzymatic.method', 'sodium', 'potassium',\n",
              "       'calcium', 'high.sensitivity.troponin'],\n",
              "      dtype='object')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 132
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "x9YDAkFvhJKp",
        "outputId": "dc674388-ace8-44d2-c0eb-e3071919f668"
      },
      "source": [
        "len(df.columns)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "27"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 133
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 437
        },
        "id": "WbgduwdgOP-z",
        "outputId": "63071fa3-bbde-465c-a090-87083501a309"
      },
      "source": [
        "sns.distplot(df['pulse'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/seaborn/distributions.py:2551: FutureWarning: `distplot` is a deprecated function and will be removed in a future version. Please adapt your code to use either `displot` (a figure-level function with similar flexibility) or `histplot` (an axes-level function for histograms).\n",
            "  warnings.warn(msg, FutureWarning)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fcafb60ef60>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 233
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oT2lIahHPgYE"
      },
      "source": [
        "Capping Brain Natriuretic Peptide at 4000 because above it doesn't relate to heart conditions and are same. citation provided"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9qBZ5HtZPpvJ"
      },
      "source": [
        "df['brain.natriuretic.peptide'].clip(upper=3000,inplace=True)\n",
        "df['creatinine.enzymatic.method'].clip(upper=600,inplace=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eQZAVoZcNl_i"
      },
      "source": [
        "Creatine Kinase above 1000: Refers to idiopathic inflammatory myopathy (IIM) hence we can cap the values to 1000.\n",
        "Hence to avoid the ouliers capped to 1000."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XO_6O2KlLZdw"
      },
      "source": [
        "df['creatine.kinase'].clip(upper=1000,inplace=True)\n",
        "df['high.sensitivity.troponin'].clip(lower = 0.1,inplace=True)\n",
        "# To prevent the division by 0"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TmiR6bcgcCuD"
      },
      "source": [
        "Converting all the numarical distributions to log normals"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "49XKeDtzcGY3"
      },
      "source": [
        "# centering the normal distributions by converting them to log\n",
        "for each in df.columns:\n",
        "    if max(df[each]) > 10 and each != 'LVEF':\n",
        "        df[each] = np.log(df[each])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CsnZmNsFk9wD"
      },
      "source": [
        "First setting cutoff of LVEF : 50 "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "phTqA52jjITo"
      },
      "source": [
        "dfef1 = df.copy()\n",
        "dfef2 = df.copy()\n",
        "dfef1['LVEF'].clip(40,41,inplace = True)\n",
        "dfef1['LVEF'].replace(to_replace=40,value=int(0),inplace=True)\n",
        "dfef1['LVEF'].replace(to_replace=41,value=int(1),inplace=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "u7kB2GVMjp0i"
      },
      "source": [
        "# Converting to a 3 class encoding\n",
        "'''\n",
        "dfef = df.copy()\n",
        "ejection_fraction = dfef['LVEF'].to_numpy()\n",
        "for i in range(len(ejection_fraction)):\n",
        "    if ejection_fraction[i] < 41 :\n",
        "        ejection_fraction[i] = -1\n",
        "    elif ejection_fraction[i] > 40 and ejection_fraction[i] < 50:\n",
        "        ejection_fraction[i] = 0\n",
        "    else:\n",
        "        ejection_fraction[i] = 1\n",
        "\n",
        "dfef1['LVEF'] = ejection_fraction\n",
        "'''\n",
        "# 3 class case did not work !!!"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "c1kEA0sslPg8"
      },
      "source": [
        "df = pd.read_csv(\"/content/07_with_LVEF_sodium_HST_BNP.csv\")\n",
        "df.drop(columns=['Unnamed: 0','map','diabetes','type.of.heart.failure','liver.disease','congestive.heart.failure','acute.renal.failure','moderate.to.severe.chronic.kidney.disease','dementia','peptic.ulcer.disease','hemiplegia','connective.tissue.disease','leukemia','AIDS','weight','height'],inplace=True)\n",
        "df['brain.natriuretic.peptide'].clip(upper=3000,inplace=True)\n",
        "df['creatinine.enzymatic.method'].clip(upper=600,inplace=True)\n",
        "\n",
        "df['creatine.kinase'].clip(upper=1000,inplace=True)\n",
        "df['high.sensitivity.troponin'].clip(lower = 0.1,inplace=True)\n",
        "# To prevent the division by 0\n",
        "\n",
        "df_og = df.copy()\n",
        "# centering the normal distributions by converting them to log\n",
        "for each in df.columns:\n",
        "    if max(df[each]) > 10 and each != 'LVEF':\n",
        "        df[each] = np.log(df[each])\n",
        "\n",
        "dfef1 = df.copy()\n",
        "dfef2 = df.copy()\n",
        "dfef3 = df.copy()\n",
        "\n",
        "dfef1['LVEF'].clip(40,41,inplace = True)\n",
        "dfef1['LVEF'].replace(to_replace=40,value=int(0),inplace=True)\n",
        "dfef1['LVEF'].replace(to_replace=41,value=int(1),inplace=True)\n",
        "\n",
        "dfef3['LVEF'].clip(40,50,inplace = True)\n",
        "dfef3['LVEF'].replace(to_replace=40,value=int(0),inplace=True)\n",
        "dfef3['LVEF'].replace(to_replace=50,value=int(1),inplace=True)\n",
        "\n",
        "Y = dfef1['LVEF'].to_numpy()\n",
        "X = pd.DataFrame(dfef1.drop(columns = ['LVEF'])).to_numpy()\n",
        "X = scaling_X(X)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "G3iII6blmlRy"
      },
      "source": [
        "from sklearn.model_selection import cross_validate\n",
        "x_train, x_test, y_train, y_test = train_test_split(X,Y,test_size = 0.2)\n",
        "scaler = preprocessing.StandardScaler().fit(x_train)\n",
        "x_train = scaler.transform(x_train)\n",
        "scaler = preprocessing.StandardScaler().fit(x_test)\n",
        "x_test = scaler.transform(x_test)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "578K-uEMT0Cb"
      },
      "source": [
        "# Robust Scaler\n",
        "x_train, x_test, y_train, y_test = train_test_split(X,Y,test_size = 0.2)\n",
        "scaler = preprocessing.RobustScaler().fit(x_train)\n",
        "x_train = scaler.transform(x_train)\n",
        "scaler = preprocessing.RobustScaler().fit(x_test)\n",
        "x_test = scaler.transform(x_test)\n",
        "\n",
        "sscaler = preprocessing.RobustScaler().fit(X)\n",
        "X = sscaler.transform(X)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HjMFXbI7myHh",
        "outputId": "56d8bf28-7dd1-497d-c758-067bb7501f46"
      },
      "source": [
        "clf = RandomForestClassifier(n_estimators=50000,min_samples_leaf=9,n_jobs=-1)\n",
        "clf.fit(x_train,y_train)\n",
        "y_predicted = clf.predict(x_test)\n",
        "y_train_pred = clf.predict(x_train)\n",
        "\n",
        "pred_prob = clf.predict_proba(x_test)\n",
        "print(\"Accuracy Test %0.2f\" %(accuracy_score(y_test,y_predicted)*100))\n",
        "print(\"Accuracy Train %0.2f\" %(accuracy_score(y_train,y_train_pred)*100))\n",
        "print(\"ROC-AUC Score %0.2f\" %(roc_auc_score(y_test,pred_prob[:,1])*100))\n",
        "print(\"Recall Score %0.2f\" %(recall_score(y_test,y_predicted)*100))\n",
        "print(\"Precision Score %0.2f\" %(precision_score(y_test,y_predicted)*100))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Accuracy Test 81.63\n",
            "Accuracy Train 88.43\n",
            "ROC-AUC Score 92.41\n",
            "Recall Score 97.33\n",
            "Precision Score 82.02\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hTZN-CCPny2R",
        "outputId": "fdd11b7b-4326-4e3b-da32-f16971de118a"
      },
      "source": [
        "cv_rf = cross_validate(clf,X,Y,scoring=('accuracy','roc_auc','recall','precision'),cv = 10,n_jobs=-1)\n",
        "print(\"Average CV Accuracy Test \\t\",(cv_rf['test_accuracy'].mean()*100))\n",
        "print(\"Average CV ROC-AUC Score \\t\",(cv_rf['test_roc_auc'].mean()*100))\n",
        "print(\"Average CV Recall Score \\t\",(cv_rf['test_recall'].mean()*100))\n",
        "print(\"Average CV Precision Score \\t\",(cv_rf['test_precision'].mean()*100))\n",
        "print(\"-----------------------------\")\n",
        "print(\"Best Accuracy CV : \\t\\t\", max(cv_rf['test_accuracy']*100))\n",
        "print(\"Best ROC CV : \\t\\t\\t\", max(cv_rf['test_roc_auc']*100))\n",
        "print(\"Best Recall CV : \\t\\t\", max(cv_rf['test_recall']*100))\n",
        "print(\"Best Precision CV : \\t\\t\", max(cv_rf['test_precision']*100))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t 80.06802721088435\n",
            "Average CV ROC-AUC Score \t 85.5034699771542\n",
            "Average CV Recall Score \t 96.23044096728309\n",
            "Average CV Precision Score \t 81.20511339014878\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t 85.71428571428571\n",
            "Best ROC CV : \t\t\t 92.56756756756758\n",
            "Best Recall CV : \t\t 100.0\n",
            "Best Precision CV : \t\t 84.0909090909091\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "id": "_KJej_9qnIcW",
        "outputId": "5cc0e666-260e-4f35-dab7-86babfb0d1d0"
      },
      "source": [
        "plt.bar(range(len(clf.feature_importances_)),clf.feature_importances_)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<BarContainer object of 26 artists>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 129
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EJUWGtWYlxhe",
        "outputId": "e1ee2981-a921-40eb-87c4-7680c269e532"
      },
      "source": [
        "from xgboost import XGBClassifier\n",
        "xgb = XGBClassifier(n_estimators=5000)\n",
        "xgb.fit(x_train,y_train)\n",
        "y_predicted = xgb.predict(x_test)\n",
        "y_train_pred = xgb.predict(x_train)\n",
        "\n",
        "pred_prob = xgb.predict_proba(x_test)\n",
        "print(\"Accuracy Test %0.2f\" %(accuracy_score(y_test,y_predicted)*100))\n",
        "print(\"Accuracy Train %0.2f\" %(accuracy_score(y_train,y_train_pred)*100))\n",
        "print(\"ROC-AUC Score %0.2f\" %(roc_auc_score(y_test,pred_prob[:,1])*100))\n",
        "print(\"Recall Score %0.2f\" %(recall_score(y_test,y_predicted)*100))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Accuracy Test 75.51\n",
            "Accuracy Train 100.00\n",
            "ROC-AUC Score 85.84\n",
            "Recall Score 95.52\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jILsSe8toMhz",
        "outputId": "d39356fd-44de-480d-9011-e88596a12347"
      },
      "source": [
        "cv_rf = cross_validate(xgb,X,Y,scoring=('accuracy','roc_auc','recall','precision'),cv = 10)\n",
        "print(\"Average CV Accuracy Test \\t\",(cv_rf['test_accuracy'].mean()*100))\n",
        "print(\"Average CV ROC-AUC Score \\t\",(cv_rf['test_roc_auc'].mean()*100))\n",
        "print(\"Average CV Recall Score \\t\",(cv_rf['test_recall'].mean()*100))\n",
        "print(\"Average CV Precision Score \\t\",(cv_rf['test_precision'].mean()*100))\n",
        "print(\"------------------------------\")\n",
        "print(\"Best Accuracy CV : \\t\\t\", max(cv_rf['test_accuracy']*100))\n",
        "print(\"Best ROC CV : \\t\\t\\t\", max(cv_rf['test_roc_auc']*100))\n",
        "print(\"Best Recall CV : \\t\\t\", max(cv_rf['test_recall']*100))\n",
        "print(\"Best Precision CV : \\t\\t\", max(cv_rf['test_precision']*100))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t 81.70918367346938\n",
            "Average CV ROC-AUC Score \t 84.68317599896548\n",
            "Average CV Recall Score \t 91.67852062588905\n",
            "Average CV Precision Score \t 85.54347743240032\n",
            "------------------------------\n",
            "Best Accuracy CV : \t\t 91.83673469387756\n",
            "Best ROC CV : \t\t\t 91.8918918918919\n",
            "Best Recall CV : \t\t 97.2972972972973\n",
            "Best Precision CV : \t\t 92.3076923076923\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "id": "-NxGF0Uwsbd0",
        "outputId": "0ab20d5a-6d5d-409c-89d5-6008576c5263"
      },
      "source": [
        "plt.bar(range(len(xgb.feature_importances_)),xgb.feature_importances_)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<BarContainer object of 26 artists>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 131
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "OpKgDzwpEnpl",
        "outputId": "7b06be5f-72db-400c-d77e-571a69b98fba"
      },
      "source": [
        "# KNN classifier\n",
        "from sklearn.neighbors import KNeighborsClassifier\n",
        "knn = KNeighborsClassifier(n_neighbors=5)\n",
        "cross_validation(knn,X,Y)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t 72.90391156462584\n",
            "Average CV ROC-AUC Score \t 72.5141169878012\n",
            "Average CV Recall Score \t 86.55049786628733\n",
            "Average CV Precision Score \t 79.7958229293709\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t 79.59183673469387\n",
            "Best ROC CV : \t\t\t 82.06388206388206\n",
            "Best Recall CV : \t\t 97.36842105263158\n",
            "Best Precision CV : \t\t 86.11111111111111\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OnGk_Xqlutml"
      },
      "source": [
        "estimators = [100,200,500,750,1000,1200,1500,2000,2500,3000,4000,5000]\n",
        "accuracy = []\n",
        "roc = []\n",
        "precision = []\n",
        "recall = []\n",
        "\n",
        "for n in estimators:\n",
        "    clf = RandomForestClassifier(n_estimators=n,min_samples_leaf= 7,n_jobs=-1)\n",
        "    cv_rf = cross_validate(clf,X,Y,scoring=('accuracy','roc_auc','recall','precision'),cv = 5)\n",
        "    accuracy.append(cv_rf['test_accuracy'].mean()*100)\n",
        "    roc.append(cv_rf['test_roc_auc'].mean()*100)\n",
        "    recall.append(cv_rf['test_recall'].mean()*100)\n",
        "    precision.append(cv_rf['test_precision'].mean()*100)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 366
        },
        "id": "Sq5zL4CCw8ue",
        "outputId": "9f8f3fb8-4017-4aad-b7c4-c8abef432362"
      },
      "source": [
        "plt.plot(estimators,accuracy,c='b',marker='x')\n",
        "plt.plot(estimators,recall,c='r',marker='o')\n",
        "plt.plot(estimators,roc,c='g',marker='d')\n",
        "plt.plot(estimators,precision,c='y',marker='.')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[<matplotlib.lines.Line2D at 0x7faca8555ac8>]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 155
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yycPgiZMjyPr"
      },
      "source": [
        "Stacking all the classifiers until now"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0fXAp-Ahd3y7",
        "outputId": "36b1c48a-3511-4b45-dab2-07b681c526ed"
      },
      "source": [
        "from sklearn.ensemble import StackingClassifier\n",
        "\n",
        "classifiers = [ ('rf',RandomForestClassifier(n_estimators=500,min_samples_leaf=7,n_jobs=-1)),\n",
        "                ('knn',KNeighborsClassifier(n_neighbors=9)),\n",
        "                ('xgb',XGBClassifier(n_estimators=1000)),\n",
        "                ('xgbrf',XGBRFClassifier(n_estimators=500,num_parallel_tree=500,num_boost_round=10))\n",
        "                ]\n",
        "final_estimator = RandomForestClassifier(n_estimators=500,min_samples_leaf=7,n_jobs=-1)\n",
        "stack_clf = StackingClassifier(estimators=classifiers,final_estimator = final_estimator, n_jobs=-1)\n",
        "\n",
        "cross_validation(stack_clf,X,Y)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t79.50\n",
            "Average CV ROC-AUC Score \t85.43\n",
            "Average CV Recall Score \t77.55\n",
            "Average CV Precision Score \t83.37\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t87.50\n",
            "Best ROC CV : \t\t\t91.08\n",
            "Best Recall CV : \t\t92.31\n",
            "Best Precision CV : \t\t91.30\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lt7QlbYTkPSz"
      },
      "source": [
        "Best Final Estimator : RandomForest\n",
        "\n",
        "For StackingClassifier\n",
        "\n",
        "```\n",
        "Average CV Accuracy Test \t  80.27\n",
        "Average CV ROC-AUC Score \t  81.67\n",
        "Average CV Recall Score \t   90.32\n",
        "Average CV Precision Score \t85.01\n",
        "------------------------------\n",
        "Best Accuracy CV  \t\t     89.79\n",
        "Best ROC CV \t\t           88.82\n",
        "Best Recall CV  \t\t       97.36\n",
        "Best Precision CV    \t\t  90.00\n",
        "```\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bnunzPqKd4C9"
      },
      "source": [
        "Train a classifier on HFrEF and HFpEF. Then predict the HFmrEF and train a final classifier.\n",
        "Even through HFmrEF is a different class we can predict that which one of the main class they are inclining towards. \n",
        "The prognosis depends on the two main classes.\n",
        "\n",
        "Hence, finally we train the calssifier for the two classess which is capable of handling the HFmrEF cases."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AZmolFa-kOjX"
      },
      "source": [
        "dfef2['LVEF'].clip(40,50,inplace = True)\n",
        "dfef2['LVEF'].replace(to_replace=40,value=int(0),inplace=True)\n",
        "dfef2['LVEF'].replace(to_replace=50,value=int(1),inplace=True)\n",
        "# index of points with LVEF 41-49\n",
        "mref_index = dfef2.loc[dfef2['LVEF'] >1].index\n",
        "#selecting the rows having EF other than 0 or 1\n",
        "HFmrEF_data = dfef2.loc[dfef2.LVEF >1]\n",
        "dfef2 = dfef2.drop(index=mref_index)  \n",
        "Y_pr = dfef2['LVEF'].to_numpy().astype(int)\n",
        "X_pr = pd.DataFrame(dfef2.drop(columns = ['LVEF'])).to_numpy()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tPf3VwC5vu-0",
        "outputId": "96598ac7-5ac9-4404-d3b7-dcfd22d71cbb"
      },
      "source": [
        "cross_validation(RandomForestClassifier(n_estimators=5000),X_pr,Y_pr)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t84.38\n",
            "Average CV ROC-AUC Score \t93.22\n",
            "Average CV Recall Score \t92.04\n",
            "Average CV Precision Score \t86.77\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t92.11\n",
            "Best ROC CV : \t\t\t97.20\n",
            "Best Recall CV : \t\t96.15\n",
            "Best Precision CV : \t\t96.00\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DRR3eNk6ysRy"
      },
      "source": [
        "Y_mr = HFmrEF_data.LVEF.to_numpy().astype(int)\n",
        "X_mr = pd.DataFrame(HFmrEF_data.drop(columns=['LVEF'])).to_numpy()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1cdB69-Lx7xk"
      },
      "source": [
        "rf_clf = RandomForestClassifier(n_estimators=5000)\n",
        "rf_clf.fit(X,Y)\n",
        "y_hfmref = rf_clf.predict(X_mr)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bq2cXSptdwUs"
      },
      "source": [
        "We have predicted the calsses of HFmrEF : \n",
        "as HFrEF and HFpEF\n",
        "\n",
        "The values were classified based on the model trained on instances of EF < 40 and EF > 50\n",
        "\n",
        "Storing the in the dataframe."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4rKhYqD2ewxz",
        "outputId": "9869b45b-4dc8-4302-941c-730aba57bd75"
      },
      "source": [
        "mref_index"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Int64Index([ 11,  12,  13,  15,  22,  26,  28,  29,  36,  43,\n",
              "            ...\n",
              "            448, 453, 454, 465, 471, 472, 474, 482, 483, 486],\n",
              "           dtype='int64', length=109)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 33
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "IG76jAYQkB88",
        "outputId": "ab3e55a5-825e-43c6-95aa-cf196cb6b476"
      },
      "source": [
        "df_og.loc[mref_index[0],'LVEF']"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "49.0"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 42
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mcASlAp0c3l7"
      },
      "source": [
        "df_mr = df.copy()\n",
        "df_og_mr = df_og.copy()\n",
        "df_og_mr['LVEF'].clip(40,50,inplace = True)\n",
        "df_og_mr['LVEF'].replace(to_replace=40,value=int(0),inplace=True)\n",
        "df_og_mr['LVEF'].replace(to_replace=50,value=int(1),inplace=True)\n",
        "\n",
        "df_mr['LVEF'].clip(40,50,inplace = True)\n",
        "df_mr['LVEF'].replace(to_replace=40,value=int(0),inplace=True)\n",
        "df_mr['LVEF'].replace(to_replace=50,value=int(1),inplace=True)\n",
        "\n",
        "for index,label in zip(mref_index,y_hfmref):\n",
        "    df_mr.loc[index,'LVEF'] =  label\n",
        "    df_og_mr.loc[index,'LVEF'] =  label"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_QaZgetBo31K"
      },
      "source": [
        "df_og_mr.to_csv(\"HFmrEF_classified_LVEF_Labels.csv\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dny8YqrHp_4r"
      },
      "source": [
        "df_mr = pd.read_csv(\"/content/HFmrEF_classified_LVEF_Labels.csv\")\n",
        "df_mr.drop(columns=['Unnamed: 0','map','diabetes'],inplace=True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jUCdFWNMpI9B"
      },
      "source": [
        "Y_mr = df_mr.LVEF.to_numpy().astype(int)\n",
        "X_mr = pd.DataFrame(df_mr.drop(columns=['LVEF'])).to_numpy()\n",
        "X_mr = scaling_X(X_mr)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "L5h-f6AynK3F"
      },
      "source": [
        "Running models on HFmrEF to 2 class converted data."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Xylj5UMynSbd",
        "outputId": "3cafc25c-6d68-44ff-820a-5e68c928bcfc"
      },
      "source": [
        "rfmr = RandomForestClassifier(n_estimators= 500,min_samples_leaf=15)\n",
        "rfmr.fit(X,Y)\n",
        "cross_validation(rfmr,X,Y)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t79.26\n",
            "Average CV ROC-AUC Score \t85.48\n",
            "Average CV Recall Score \t97.84\n",
            "Average CV Precision Score \t79.67\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t83.67\n",
            "Best ROC CV : \t\t\t92.34\n",
            "Best Recall CV : \t\t100.00\n",
            "Best Precision CV : \t\t82.22\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TRqpEh0V0MX0"
      },
      "source": [
        "Random Forest\n",
        "\n",
        "LVEF cutoff : 40\n",
        "\n",
        "\n",
        "```\n",
        "Average CV Accuracy Test \t80.08\n",
        "Average CV ROC-AUC Score \t85.88\n",
        "Average CV Recall Score \t 97.84\n",
        "Average CV Precision Score   80.35\n",
        "-----------------------------\n",
        "Best Accuracy CV : \t\t  83.67\n",
        "Best ROC CV : \t\t\t   92.14\n",
        "Best Recall CV : \t\t    100.0\n",
        "Best Precision CV : \t\t 82.61\n",
        "```\n",
        "HFmrEF distributed classified into 2 classes\n",
        "\n",
        "\n",
        "\n",
        "```\n",
        "Average CV Accuracy Test \t83.15\n",
        "Average CV ROC-AUC Score \t93.85\n",
        "Average CV Recall Score \t 95.30\n",
        "Average CV Precision Score   82.22\n",
        "```\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WqeoqbFinKDq",
        "outputId": "1d8ba344-83cc-4bf5-fe32-bb62825b07bc"
      },
      "source": [
        "from sklearn.ensemble import StackingClassifier\n",
        "classifiers = [ ('rf',RandomForestClassifier(n_estimators=500,min_samples_leaf=13,n_jobs=-1)),\n",
        "                ('knn',KNeighborsClassifier(n_neighbors=9)),\n",
        "                ('xgb',XGBClassifier(n_estimators=1000)),\n",
        "                ('xgbrf',XGBRFClassifier(n_estimators=500,num_parallel_tree=500,num_boost_round=10))\n",
        "                ]\n",
        "final_estimator = RandomForestClassifier(n_estimators=500,min_samples_leaf=7,n_jobs=-1)\n",
        "stack_clf = StackingClassifier(estimators=classifiers,final_estimator = final_estimator, n_jobs=-1)\n",
        "\n",
        "cross_validation(stack_clf,X,Y)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t80.68\n",
            "Average CV ROC-AUC Score \t84.52\n",
            "Average CV Recall Score \t91.96\n",
            "Average CV Precision Score \t84.45\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t87.50\n",
            "Best ROC CV : \t\t\t90.91\n",
            "Best Recall CV : \t\t100.00\n",
            "Best Precision CV : \t\t91.67\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xlgmTWUbdQiM"
      },
      "source": [
        "Stacking Classifier:\n",
        "\n",
        "When HFmrEF is classified into 2 classes and model is trained on resultant data.\n",
        "\n",
        "```\n",
        "Average CV Accuracy Test \t 88.095\n",
        "Average CV ROC-AUC Score \t 94.013\n",
        "Average CV Recall Score \t 90.282\n",
        "Average CV Precision Score \t 91.712\n",
        "-----------------------------\n",
        "Best Accuracy CV : \t\t 93.877\n",
        "Best ROC CV : \t\t\t 98.292\n",
        "Best Recall CV : \t\t 96.875\n",
        "Best Precision CV : \t\t 100.0\n",
        "```\n",
        "\n",
        "Otherwise with 40 cutoff : \n",
        "\n",
        "\n",
        "```\n",
        "Average CV Accuracy Test \t 80.684\n",
        "Average CV ROC-AUC Score \t 85.317\n",
        "Average CV Recall Score \t 97.034\n",
        "Average CV Precision Score \t 81.336\n",
        "-----------------------------\n",
        "Best Accuracy CV : \t\t 85.71\n",
        "Best ROC CV : \t\t\t 92.34\n",
        "Best Recall CV : \t\t 100.0\n",
        "Best Precision CV : \t\t 84.44\n",
        "```\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "TQeQOvSJtLa4",
        "outputId": "8395cef4-3e7f-43e9-fabf-4cc38a2a7027"
      },
      "source": [
        "from lightgbm import LGBMClassifier\n",
        "lgb = LGBMClassifier(learning_rate=0.5,num_iterations = 5000)\n",
        "cross_validation(lgb,X,Y)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t 81.09268707482993\n",
            "Average CV ROC-AUC Score \t 85.42264752791068\n",
            "Average CV Recall Score \t 90.86059743954482\n",
            "Average CV Precision Score \t 85.52977724030356\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t 85.71428571428571\n",
            "Best ROC CV : \t\t\t 92.7927927927928\n",
            "Best Recall CV : \t\t 97.36842105263158\n",
            "Best Precision CV : \t\t 91.66666666666666\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "T9CsGk2vsCw7",
        "outputId": "4fa8d512-8d3f-4f30-ca54-e45abd300e50"
      },
      "source": [
        "from sklearn.ensemble import VotingClassifier\n",
        "from lightgbm import LGBMClassifier\n",
        "classifiers = [ ('rf',RandomForestClassifier(n_estimators=500,min_samples_leaf=7,n_jobs=-1)),\n",
        "                ('knn',KNeighborsClassifier(n_neighbors=9)),\n",
        "                ('xgb',XGBClassifier(n_estimators=1000)),\n",
        "                ('xgbrf',XGBRFClassifier(n_estimators=500,num_parallel_tree=500,num_boost_round=10)),\n",
        "                ('lgb',LGBMClassifier(learning_rate=0.5,num_iterations = 1500))\n",
        "                ]\n",
        "voting_clf = VotingClassifier(estimators=classifiers,voting='soft', n_jobs=-1)\n",
        "cross_validation(voting_clf,X_mr,Y_mr)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t87.88\n",
            "Average CV ROC-AUC Score \t94.66\n",
            "Average CV Recall Score \t91.54\n",
            "Average CV Precision Score \t90.24\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t93.88\n",
            "Best ROC CV : \t\t\t98.35\n",
            "Best Recall CV : \t\t96.88\n",
            "Best Precision CV : \t\t100.00\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SLA0QdXbvsyU"
      },
      "source": [
        "Voting Classifer : Hard\n",
        "\n",
        "LVEF cutoff : 40\n",
        "\n",
        "\n",
        "```\n",
        "Average CV Accuracy Test \t 81.30\n",
        "Average CV Recall Score \t  95.68\n",
        "Average CV Precision Score    82.61\n",
        "-----------------------------\n",
        "Best Accuracy CV : \t\t   87.75\n",
        "Best Recall CV : \t\t     100.0\n",
        "Best Precision CV : \t\t  86.05\n",
        "\n",
        "```\n",
        "\n",
        "HFmrEF classfied into 2 classes\n",
        "\n",
        "Voting : Soft \n",
        "\n",
        "```\n",
        "Average CV Accuracy Test \t 88.29\n",
        "Average CV Recall Score \t  91.85\n",
        "Average CV Precision Score    90.60\n",
        "-----------------------------\n",
        "Best Accuracy CV : \t\t   93.87\n",
        "Best Recall CV : \t\t     96.87\n",
        "Best Precision CV : \t\t  100.0\n",
        "```\n",
        "Voting : Soft\n",
        "\n",
        "\n",
        "```\n",
        "Average CV Accuracy Test \t87.88\n",
        "Average CV ROC-AUC Score \t94.66\n",
        "Average CV Recall Score \t 91.54\n",
        "Average CV Precision Score   90.24\n",
        "-----------------------------\n",
        "Best Accuracy CV : \t\t  93.88\n",
        "Best ROC CV : \t\t\t   98.35\n",
        "Best Recall CV : \t\t    96.88\n",
        "Best Precision CV : \t\t 100.0\n",
        "```\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZQZbE6sDwvkZ",
        "outputId": "686fc22e-561a-4998-d52f-d18eb2aa2b4b"
      },
      "source": [
        "df_mr.columns\n",
        "features = [0,2,3,14,15,18,19]\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Index(['pulse', 'respiration', 'systolic.blood.pressure',\n",
              "       'diastolic.blood.pressure', 'BMI',\n",
              "       'NYHA.cardiac.function.classification', 'Killip.grade',\n",
              "       'myocardial.infarction', 'peripheral.vascular.disease',\n",
              "       'cerebrovascular.disease', 'Chronic.obstructive.pulmonary.disease',\n",
              "       'CCI.score', 'type.II.respiratory.failure', 'fio2', 'LVEF',\n",
              "       'left.ventricular.end.diastolic.diameter.LV',\n",
              "       'brain.natriuretic.peptide', 'cystatin', 'creatine.kinase',\n",
              "       'cholesterol', 'creatinine.enzymatic.method', 'sodium', 'potassium',\n",
              "       'calcium', 'high.sensitivity.troponin'],\n",
              "      dtype='object')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 33
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 381
        },
        "id": "yOAWS54r4faI",
        "outputId": "50bc2f6f-b3b7-41df-9bd7-babd48801598"
      },
      "source": [
        "plt.bar(range(len(rfmr.feature_importances_)),rfmr.feature_importances_)\n",
        "plt.title(\"Feature importances for RF on HFmrEF classified\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Feature importances for RF on HFmrEF classified')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 59
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 381
        },
        "id": "NwVdpOf57IZ8",
        "outputId": "32f35051-1b25-4024-b389-99adbc5154eb"
      },
      "source": [
        "plt.bar(range(len(rfmr.feature_importances_)),rfmr.feature_importances_)\n",
        "plt.title(\"Feature importances for RF on cutoff of 40\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Feature importances for RF on cutoff of 40')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 71
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Z98XSyv57YLL"
      },
      "source": [
        "Taking only :\n",
        "features = [0,2,3,4,14,15,18,19]\n",
        "\n",
        "* 0 : Pulse\n",
        "* 2 : SBP\n",
        "* 3 : DBP\n",
        "* 4 : BMI\n",
        "* 14: LV end diastolic Diameter\n",
        "* 15: BNP\n",
        "* 18: Cholesterol\n",
        "* 19: Creaatinine Enzymatic Method"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6pV9J9mw7wnT"
      },
      "source": [
        "df_reduced = dfef1.loc[:,['pulse','systolic.blood.pressure','diastolic.blood.pressure','BMI','LVEF','left.ventricular.end.diastolic.diameter.LV',\n",
        "                        'brain.natriuretic.peptide','cholesterol', 'creatinine.enzymatic.method']]\n",
        "df_mr_reduced = df_mr.loc[:,['pulse','systolic.blood.pressure','diastolic.blood.pressure','BMI','LVEF','left.ventricular.end.diastolic.diameter.LV',\n",
        "                        'brain.natriuretic.peptide','cholesterol', 'creatinine.enzymatic.method']]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TNk7mCSi8pvH"
      },
      "source": [
        "Y_r = df_reduced.LVEF.to_numpy().astype(int)\n",
        "X_r = pd.DataFrame(df_reduced.drop(columns=['LVEF'])).to_numpy()\n",
        "X_r = scaling_X(X_r)\n",
        "\n",
        "Y_mr_r = df_mr_reduced.LVEF.to_numpy().astype(int)\n",
        "X_mr_r = pd.DataFrame(df_mr_reduced.drop(columns=['LVEF'])).to_numpy()\n",
        "X_mr_r = scaling_X(X_mr_r)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xrcl1GD79wj9",
        "outputId": "e46ce3b2-e310-4a01-9ead-7d8eef0c4265"
      },
      "source": [
        "rfr = RandomForestClassifier(n_estimators=500,min_samples_leaf=13)\n",
        "rfr.fit(X_mr_r,Y_mr_r)\n",
        "cross_validation(rfr,X_mr_r,Y_mr_r)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t88.10\n",
            "Average CV ROC-AUC Score \t94.68\n",
            "Average CV Recall Score \t91.85\n",
            "Average CV Precision Score \t90.43\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t95.92\n",
            "Best ROC CV : \t\t\t98.53\n",
            "Best Recall CV : \t\t96.88\n",
            "Best Precision CV : \t\t100.00\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pcH3zMnH7wcG"
      },
      "source": [
        "RF accuracy on reduced features\n",
        "cutoff 40\n",
        "\n",
        "```\n",
        "Average CV Accuracy Test \t80.28\n",
        "Average CV ROC-AUC Score \t86.66\n",
        "Average CV Recall Score \t 95.16\n",
        "Average CV Precision Score   81.95\n",
        "```\n",
        "\n",
        "on HFmrRF classified to other classes\n",
        "\n",
        "\n",
        "\n",
        "```\n",
        "Average CV Accuracy Test \t88.31\n",
        "Average CV ROC-AUC Score \t94.73\n",
        "Average CV Recall Score \t 92.18\n",
        "Average CV Precision Score   90.47\n",
        "```\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 436
        },
        "id": "5gWO6dwp5PfZ",
        "outputId": "e940041e-a0b8-4a54-b695-39d51b1a9236"
      },
      "source": [
        "plt.style.use(\"seaborn\")\n",
        "x_label = ['pulse','Systolic BP','Diastolic BP','BMI','Left.ventricular.end.diastolic.diameter','Brain.natriuretic.peptide','Cholesterol','Creatinine Enzymatic']\n",
        "plt.barh(x_label,rfr.feature_importances_)\n",
        "plt.title(\"Feature Importance for RFClassifier (HFmrEF classified)\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Feature Importance for RFClassifier (HFmrEF classified)')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 63
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 640x440 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zr3JXbgP9HNC"
      },
      "source": [
        "x_label = ['pulse','Systolic BP','Diastolic BP','BMI','Left.ventricular.end.diastolic.diameter','brain.natriuretic.peptide','Cholesterol','Brain Natriuretic Peptide']\n",
        "plt.barh(x_label,rfr.feature_importances_)\n",
        "plt.title(\"Feature Importance for RFClassifier\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kbSu34e0WnYH"
      },
      "source": [
        "plt.style.use('classic')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nAj63VZuQYyZ"
      },
      "source": [
        "num = {0:'pulse',1:'systolic.blood.pressure',3:'BMI',4:'left.ventricular.end.diastolic.diameter.LV',\n",
        "                        5:'brain.natriuretic.peptide',6:'cholesterol',7: 'creatinine.enzymatic.method'}\n",
        "number = [0,1,3,4,5,6,7]\n",
        "def plot_decision_boundaries(X, y,name, model_class, **model_params):\n",
        "    try:\n",
        "        X = np.array(X)\n",
        "        y = np.array(y).flatten()\n",
        "    except:\n",
        "        print(\"Coercing input data to NumPy arrays failed\")\n",
        "    # Reduces to the first two columns of data\n",
        "    for n in range(0,len(number)):\n",
        "        # print(x)\n",
        "        for nn in range(n+1,len(number)):\n",
        "            # XX = X[:,[n,nn]].copy()\n",
        "            reduced_data = X[:, [number[n],number[nn]]]\n",
        "            # Instantiate the model object\n",
        "            model = model_class(**model_params)\n",
        "            # Fits the model with the reduced data\n",
        "            model.fit(reduced_data, y)\n",
        "\n",
        "            # Step size of the mesh. Decrease to increase the quality of the VQ.\n",
        "            h = .02     # point in the mesh [x_min, m_max]x[y_min, y_max].    \n",
        "\n",
        "            # Plot the decision boundary. For that, we will assign a color to each\n",
        "            x_min, x_max = reduced_data[:, 0].min() - 1, reduced_data[:, 0].max() + 1\n",
        "            y_min, y_max = reduced_data[:, 1].min() - 1, reduced_data[:, 1].max() + 1\n",
        "            # Meshgrid creation\n",
        "            xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))\n",
        "\n",
        "            # Obtain labels for each point in mesh using the model.\n",
        "            Z = model.predict(np.c_[xx.ravel(), yy.ravel()])    \n",
        "\n",
        "            x_min, x_max = reduced_data[:, 0].min() - 1, reduced_data[:, 0].max() + 1\n",
        "            y_min, y_max = reduced_data[:, 1].min() - 1, reduced_data[:, 1].max() + 1\n",
        "            xx, yy = np.meshgrid(np.arange(x_min, x_max, 0.1),\n",
        "                                np.arange(y_min, y_max, 0.1))\n",
        "\n",
        "            # Predictions to obtain the classification results\n",
        "            Z = model.predict(np.c_[xx.ravel(), yy.ravel()]).reshape(xx.shape)\n",
        "\n",
        "            # Plotting\n",
        "            plt.figure(figsize=(12,7))\n",
        "            plt.contourf(xx, yy, Z, alpha=0.25)\n",
        "            plt.scatter(reduced_data[:, 0], reduced_data[:, 1], c=y, alpha=0.8)\n",
        "            plt.xlabel(\"Feature-1\",fontsize=5)\n",
        "            plt.ylabel(\"Feature-2\",fontsize=5)\n",
        "            plt.xticks(fontsize=8)\n",
        "            plt.yticks(fontsize=8)\n",
        "            plt.title(f'{model_class}-{num[number[n]]} vs {num[number[nn]]}')\n",
        "            #plt.savefig(f'four_features/{num[number[n]]} vs {num[number[nn]]}___{name}__FOUR.png')\n",
        "            # return plt"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BZghsXgdQevc"
      },
      "source": [
        "plot_decision_boundaries(X_mr_r,Y_mr_r,'RandomForestClassifier',RandomForestClassifier,n_estimators=500,min_samples_leaf=13)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hxKDMPT-YCX0"
      },
      "source": [
        "plot_decision_boundaries(X_r,Y_r,'RandomForestClassifier',RandomForestClassifier,n_estimators=500,min_samples_leaf=13)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wOI2ISnq0xsZ"
      },
      "source": [
        "###Extract the data of reduced features as samples might increase.\r\n",
        "\r\n",
        "Now train again for HFmrEF and cutoff with LVEF 40"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yNIRlIbRhpEE"
      },
      "source": [
        "i = -1\r\n",
        "for each in df.columns:\r\n",
        "    i+=1\r\n",
        "    if each == 'LVEF':\r\n",
        "        i-=1\r\n",
        "        continue\r\n",
        "    print(i,each)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4LAze5Rh0xIx"
      },
      "source": [
        "xgb = XGBClassifier(n_estimators = 1000,n_jobs=-1,).fit(X,Y)\r\n",
        "rf = RandomForestClassifier(n_estimators = 500,n_jobs=-1).fit(X,Y)\r\n",
        "lgb = LGBMClassifier().fit(X,Y)\r\n",
        "cat = CatBoostClassifier(iterations=1000,silent=True).fit(X,Y)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3i9kADGkgWOB"
      },
      "source": [
        "plt.bar(range(len(xgb.feature_importances_)),xgb.feature_importances_)\r\n",
        "plt.title(\"Feature importances for RF on cutoff of 40\")\r\n",
        "#cross_validation(cat,X,Y)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4jfYiyz3EJ41"
      },
      "source": [
        "Training the model on pure HFrEF and HFpEF to predict label for HFmrEF"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jpNpweEY7-Ne"
      },
      "source": [
        "df_red = pd.read_csv(\"/content/HF_Classification_Reduced_Features.csv\")\r\n",
        "\r\n",
        "#clipping the outliers\r\n",
        "df_red.drop(columns=['Unnamed: 0','Killip.grade'],inplace=True)\r\n",
        "df_red['brain.natriuretic.peptide'].clip(upper=3000,inplace=True)\r\n",
        "df_red['creatinine.enzymatic.method'].clip(upper=600,inplace=True)\r\n",
        "df_red['creatine.kinase'].clip(upper=1000,inplace=True)\r\n",
        "\r\n",
        "# Log transformation\r\n",
        "for each in df_red.columns:\r\n",
        "    if max(df_red[each]) > 10 and each != 'LVEF':\r\n",
        "        df_red[each] = np.log(df_red[each])"
      ],
      "execution_count": 120,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jtjCuDHpExrV",
        "outputId": "7665519f-ca2c-492f-a441-9e157b4555fa"
      },
      "source": [
        "df_red.columns\r\n",
        "df_red.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(495, 10)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "70fi-cpc7EQG"
      },
      "source": [
        "df_red1 = df_red.copy()\r\n",
        "df_red1['LVEF'].clip(40,50,inplace = True)\r\n",
        "df_red1['LVEF'].replace(to_replace=40,value=int(0),inplace=True)\r\n",
        "df_red1['LVEF'].replace(to_replace=50,value=int(1),inplace=True)\r\n",
        "# index of points with LVEF 41-49\r\n",
        "mref_index = df_red1.loc[df_red1['LVEF'] >1].index\r\n",
        "#selecting the rows having EF other than 0 or 1\r\n",
        "HFmrEF_data = df_red1.loc[df_red1.LVEF >1]\r\n",
        "df_red1 = df_red1.drop(index=mref_index)  \r\n",
        "Y_pr = df_red1['LVEF'].to_numpy().astype(int)\r\n",
        "X_pr = pd.DataFrame(df_red1.drop(columns = ['LVEF'])).to_numpy()"
      ],
      "execution_count": 121,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Oo2HKITTFFYQ",
        "outputId": "e27bab87-03fa-4a88-f262-01a4195c74d5"
      },
      "source": [
        "len(X_pr)"
      ],
      "execution_count": 50,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "384"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 50
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-9s12kcXEuQJ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "dd7df15d-cde4-41b2-e712-0b2d14ac1255"
      },
      "source": [
        "# Augmenting the synthetic samples to tackle class imbalance\r\n",
        "from imblearn.over_sampling import ADASYN\r\n",
        "ada = ADASYN()\r\n",
        "X_pr_aug, Y_pr_aug = ada.fit_resample(X_pr,Y_pr)"
      ],
      "execution_count": 122,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/sklearn/utils/deprecation.py:87: FutureWarning: Function safe_indexing is deprecated; safe_indexing is deprecated in version 0.22 and will be removed in version 0.24.\n",
            "  warnings.warn(msg, category=FutureWarning)\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3RG_0RklKKq2",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "3c16384b-ebe6-4e71-a41a-8161dcb2c58f"
      },
      "source": [
        "lgb = LGBMClassifier(n_estimators=1000)\r\n",
        "lgb.fit(X_pr_aug,Y_pr_aug)"
      ],
      "execution_count": 123,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "LGBMClassifier(boosting_type='gbdt', class_weight=None, colsample_bytree=1.0,\n",
              "               importance_type='split', learning_rate=0.1, max_depth=-1,\n",
              "               min_child_samples=20, min_child_weight=0.001, min_split_gain=0.0,\n",
              "               n_estimators=1000, n_jobs=-1, num_leaves=31, objective=None,\n",
              "               random_state=None, reg_alpha=0.0, reg_lambda=0.0, silent=True,\n",
              "               subsample=1.0, subsample_for_bin=200000, subsample_freq=0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 123
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "y0e1pvYZMeWS"
      },
      "source": [
        "#X_pr_aug = scaling_X(X_pr_aug,X_pr_aug)"
      ],
      "execution_count": 65,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "_z5yAGiyLv8V",
        "outputId": "29484591-b6dc-47c0-ecd0-e4a258ea5c22"
      },
      "source": [
        "cross_validation(lgb,X_pr_aug,Y_pr_aug)"
      ],
      "execution_count": 124,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t91.02\n",
            "Average CV ROC-AUC Score \t95.24\n",
            "Average CV Recall Score \t86.87\n",
            "Average CV Precision Score \t94.64\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t94.44\n",
            "Best ROC CV : \t\t\t98.49\n",
            "Best Recall CV : \t\t96.30\n",
            "Best Precision CV : \t\t100.00\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-bpHlfxMHwXe",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a3d19556-1250-4d71-c03b-d258783e054a"
      },
      "source": [
        "rfmr = RandomForestClassifier(n_estimators=1000, n_jobs=-1)\r\n",
        "rfmr.fit(X_pr_aug,Y_pr_aug)\r\n",
        "cross_validation(rfmr,X_pr_aug,Y_pr_aug)"
      ],
      "execution_count": 125,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t89.74\n",
            "Average CV ROC-AUC Score \t95.46\n",
            "Average CV Recall Score \t84.27\n",
            "Average CV Precision Score \t94.32\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t96.30\n",
            "Best ROC CV : \t\t\t99.86\n",
            "Best Recall CV : \t\t96.30\n",
            "Best Precision CV : \t\t100.00\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "N8yuI0C3AnIy"
      },
      "source": [
        "# Depricated. Replaced with augmented trained set from ADASYN\r\n",
        "rfmr = RandomForestClassifier(n_estimators=500,n_jobs=-1)\r\n",
        "rfmr.fit(X_pr,Y_pr)\r\n",
        "cross_validation(rfmr,X_pr,Y_pr)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Gf_4cl2YEZtM"
      },
      "source": [
        "plt.bar(range(len(rfmr.feature_importances_)),rfmr.feature_importances_)\r\n",
        "plt.title(\"Feature importances for RF on cutoff of 40\")"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "M0Vpd_KP7T4I"
      },
      "source": [
        "Y_mr = HFmrEF_data.LVEF.to_numpy().astype(int)\r\n",
        "X_mr = pd.DataFrame(HFmrEF_data.drop(columns=['LVEF'])).to_numpy()"
      ],
      "execution_count": 128,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PdJCppaMNKOU"
      },
      "source": [
        "X_mr = scaling_X(X_pr_aug,X_mr)"
      ],
      "execution_count": 68,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lNutI9TEJt3s"
      },
      "source": [
        "y_hfmref_rf = rfmr.predict(X_mr)\r\n",
        "y_hfmref_lgb = lgb.predict(X_mr)"
      ],
      "execution_count": 134,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "R_MiXdio71fM"
      },
      "source": [
        "# df_mr = df_red.copy()\n",
        "# df_mr['LVEF'].clip(40,50,inplace = True)\n",
        "# df_mr['LVEF'].replace(to_replace=40,value=int(0),inplace=True)\n",
        "# df_mr['LVEF'].replace(to_replace=50,value=int(1),inplace=True)\n",
        "# #y_hfmref : Predicted labels from HFrEF and HFpEF classifier\n",
        "# for index,label in zip(mref_index,y_hfmref_rf):\n",
        "#     df_mr.loc[index,'LVEF'] = label\n"
      ],
      "execution_count": 109,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "czrOtotg71fR"
      },
      "source": [
        "# Dataset with HFmrEF clasified into p and r\n",
        "# Y_mrp = df_mr.LVEF.to_numpy().astype(int)\n",
        "# X_mrp = pd.DataFrame(df_mr.drop(columns=['LVEF'])).to_numpy()\n",
        "# X_mrp = scaling_X(X_mrp,X_mrp)\n",
        "\n",
        "X_mrp = np.concatenate((X_pr_aug,X_mr))\n",
        "Y_mrp_lgb = np.concatenate((Y_pr_aug,y_hfmref_lgb))\n",
        "Y_mrp_rf = np.concatenate((Y_pr_aug,y_hfmref_rf))\n",
        "X_mrp = scaling_X(X_mrp,X_mrp)"
      ],
      "execution_count": 137,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "b9dmvayhS-ua",
        "outputId": "0eb2e03f-8b54-44ce-df5d-6c2abf1abe6c"
      },
      "source": [
        "Y_mrp.shape"
      ],
      "execution_count": 132,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(655,)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 132
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "e-hOYRoGM3S_"
      },
      "source": [
        "RF and LGBM Model with HFmrEF labelled as P and R based on selected 9 features."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EV_lcCZPPGgS"
      },
      "source": [
        "Models Trained on HFmrEF labels obtained from LBGM model\r\n",
        "\r\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vCu_oGrbTJUM",
        "outputId": "79df1b50-ccc1-4d7a-f50b-9862c5dde92f"
      },
      "source": [
        "# Final set = Aug_set +labelled\r\n",
        "cross_validation(RandomForestClassifier(n_estimators=500,n_jobs=-1)\r\n",
        "                ,X_mrp,Y_mrp_lgb)"
      ],
      "execution_count": 142,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t89.93\n",
            "Average CV ROC-AUC Score \t95.52\n",
            "Average CV Recall Score \t83.67\n",
            "Average CV Precision Score \t94.70\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t96.92\n",
            "Best ROC CV : \t\t\t98.67\n",
            "Best Recall CV : \t\t93.55\n",
            "Best Precision CV : \t\t100.00\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7G4VmJKQU9u3",
        "outputId": "db7611fc-2839-4a66-c3f8-706bf766afc3"
      },
      "source": [
        "# Final set = Aug_set +labelled\r\n",
        "cross_validation(LGBMClassifier(n_estimators=1000),X_mrp,Y_mrp_lgb)"
      ],
      "execution_count": 143,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t92.54\n",
            "Average CV ROC-AUC Score \t96.14\n",
            "Average CV Recall Score \t87.51\n",
            "Average CV Precision Score \t96.59\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t98.46\n",
            "Best ROC CV : \t\t\t99.53\n",
            "Best Recall CV : \t\t96.77\n",
            "Best Precision CV : \t\t100.00\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-78fOCCyOcU7",
        "outputId": "72d06d27-18ef-4c3b-a6bb-2e5ddd328bd0"
      },
      "source": [
        "cross_validation(RandomForestClassifier(n_estimators=500,n_jobs=-1)\r\n",
        "                ,X_mrp,Y_mrp)"
      ],
      "execution_count": 101,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t87.69\n",
            "Average CV ROC-AUC Score \t94.14\n",
            "Average CV Recall Score \t90.12\n",
            "Average CV Precision Score \t90.53\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t92.00\n",
            "Best ROC CV : \t\t\t97.85\n",
            "Best Recall CV : \t\t96.77\n",
            "Best Precision CV : \t\t96.67\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "L-0ABmzlKrZI",
        "outputId": "3e76e30a-7225-42fa-a64d-af9fd170c84b"
      },
      "source": [
        "cross_validation(LGBMClassifier(n_estimators=1000),X_mrp,Y_mrp)"
      ],
      "execution_count": 102,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t86.27\n",
            "Average CV ROC-AUC Score \t94.20\n",
            "Average CV Recall Score \t89.48\n",
            "Average CV Precision Score \t89.14\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t92.00\n",
            "Best ROC CV : \t\t\t97.49\n",
            "Best Recall CV : \t\t96.77\n",
            "Best Precision CV : \t\t96.67\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EVmoEH5_PNvJ"
      },
      "source": [
        "Models Trained on HFmrEF labels obtained from RF model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Eigy0nFIT9ni",
        "outputId": "6746f86f-0183-40a5-f273-f43dfb882c12"
      },
      "source": [
        "# Final set = Aug_set +labelled\r\n",
        "cross_validation(RandomForestClassifier(n_estimators=100,n_jobs=-1)\r\n",
        "                ,X_mrp,Y_mrp_rf)"
      ],
      "execution_count": 141,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t91.32\n",
            "Average CV ROC-AUC Score \t95.71\n",
            "Average CV Recall Score \t86.09\n",
            "Average CV Precision Score \t95.28\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t98.46\n",
            "Best ROC CV : \t\t\t99.81\n",
            "Best Recall CV : \t\t96.77\n",
            "Best Precision CV : \t\t100.00\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Kiw2Uls7PMBI",
        "outputId": "2349aeb1-0427-4bb0-d0b4-4099ae4ffb60"
      },
      "source": [
        "cross_validation(RandomForestClassifier(n_estimators=100,n_jobs=-1)\r\n",
        "                ,X_mrp,Y_mrp)"
      ],
      "execution_count": 111,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t88.31\n",
            "Average CV ROC-AUC Score \t93.94\n",
            "Average CV Recall Score \t89.31\n",
            "Average CV Precision Score \t91.82\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t95.92\n",
            "Best ROC CV : \t\t\t97.40\n",
            "Best Recall CV : \t\t96.77\n",
            "Best Precision CV : \t\t96.77\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "iTmGl5Y-Pklg",
        "outputId": "b9046692-25de-4760-98e4-5dfc88e4e49e"
      },
      "source": [
        "cross_validation(LGBMClassifier(n_estimators=100),X_mrp,Y_mrp)"
      ],
      "execution_count": 113,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Average CV Accuracy Test \t86.10\n",
            "Average CV ROC-AUC Score \t93.27\n",
            "Average CV Recall Score \t88.35\n",
            "Average CV Precision Score \t89.54\n",
            "-----------------------------\n",
            "Best Accuracy CV : \t\t93.88\n",
            "Best ROC CV : \t\t\t96.77\n",
            "Best Recall CV : \t\t96.77\n",
            "Best Precision CV : \t\t100.00\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9YJBG2s4VTqR"
      },
      "source": [
        "##Adding 3 Medicinal features : Diuretics, Vasodilators and Inotropes"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tI-YZdKiVgtd"
      },
      "source": [
        "df = pd.read_csv(\"/content/HF_Classification_reduced_with_Medicinal_features.csv\")\r\n",
        "df.drop(columns= [\"Unnamed: 0\",'Killip.grade'],inplace = True)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "blCGPz_9YE5-"
      },
      "source": [
        "dfef = df.copy()\r\n",
        "for each in dfef.columns:\r\n",
        "    if max(dfef[each]) > 10 and each != 'LVEF':\r\n",
        "        dfef[each].clip(lower=0.0000001,inplace = True)\r\n",
        "        dfef[each] = np.log(dfef[each])\r\n",
        "dfef['LVEF'].clip(40,41,inplace = True)\r\n",
        "dfef['LVEF'].replace(to_replace=40,value=int(0),inplace=True)\r\n",
        "dfef['LVEF'].replace(to_replace=41,value=int(1),inplace=True)\r\n",
        "Y_med = dfef['LVEF'].to_numpy().astype(int)\r\n",
        "X_med = pd.DataFrame(dfef.drop(columns = ['LVEF'])).to_numpy()\r\n",
        "X_med = scaling_X(X_med)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CkEwsO5oZSKK"
      },
      "source": [
        "rf = RandomForestClassifier(n_estimators=1000,n_jobs=-1)\r\n",
        "rf.fit(X_med,Y_med)\r\n",
        "#cross_validation(rf,X_med,Y_med)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 299
        },
        "id": "hNKlsCFiZpn6",
        "outputId": "c089025e-da30-453b-ed4f-bc70235f6dcb"
      },
      "source": [
        "plt.bar(range(len(rf.feature_importances_)),rf.feature_importances_)\r\n",
        "plt.title(\"Feature importances for RF on cutoff of 40\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "Text(0.5, 1.0, 'Feature importances for RF on cutoff of 40')"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "krKhqutza-37"
      },
      "source": [
        "df_red1 = df.copy()\r\n",
        "for each in df_red1.columns:\r\n",
        "    if max(df_red1[each]) > 10 and each != 'LVEF':\r\n",
        "        df_red1[each].clip(lower=0.0000001,inplace = True)\r\n",
        "        df_red1[each] = np.log(df_red1[each])\r\n",
        "df_red1['LVEF'].clip(40,50,inplace = True)\r\n",
        "df_red1['LVEF'].replace(to_replace=40,value=int(0),inplace=True)\r\n",
        "df_red1['LVEF'].replace(to_replace=50,value=int(1),inplace=True)\r\n",
        "# index of points with LVEF 41-49\r\n",
        "mref_index = df_red1.loc[df_red1['LVEF'] >1].index\r\n",
        "#selecting the rows having EF other than 0 or 1\r\n",
        "HFmrEF_data = df_red1.loc[df_red1.LVEF >1]\r\n",
        "df_red1 = df_red1.drop(index=mref_index)  \r\n",
        "Y_pr = df_red1['LVEF'].to_numpy().astype(int)\r\n",
        "X_pr = pd.DataFrame(df_red1.drop(columns = ['LVEF'])).to_numpy()"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KtfKcximk_nR"
      },
      "source": [
        "cat = CatBoostClassifier(iterations = 50000,border_count = 254).fit(X_pr,Y_pr)\r\n",
        "#cross_validation(cat,X_pr,Y_pr)"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}