W01_end_to_end_case.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Science 3 (prediction)\n",
    "\n",
    "## End to end example\n",
    "\n",
    "**Teachers:**\n",
    "\n",
    "* Supervised learning: Bart Barnard (BABA)\n",
    "\n",
    "**Data files:**\n",
    "\n",
    "* heart_failure_clinical_records_dataset.csv\n",
    "* data_description.csv\n",
    "\n",
    "\n",
    "This notebook demonstrates the steps to build a supervised machine learning model. The following steps are performed\n",
    "\n",
    "1. Data is loaded. \n",
    "2. Data is inspected\n",
    "3. Data is visualized to gain more understanding about the data\n",
    "4. Data is prepared for the model to train\n",
    "5. Model is trained on the prepared data\n",
    "6. Model is used to test the working of the model\n",
    "7. Model is evaluated\n",
    "8. Model is improved\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "\n",
    "##  Supervised learning \n",
    "\n",
    "Cardiovascular diseases kill approximately 17 million people globally every year, and they mainly exhibit as myocardial infarctions and heart failures. Heart failure occurs when the heart cannot pump enough blood to meet the needs of the body. Available electronic medical records of patients quantify symptoms, body features, and clinical laboratory test values, which can be used to perform biostatistics analysis aimed at highlighting patterns and correlations otherwise undetectable by medical doctors. Machine learning, can predict patients’ survival from their data and can individuate the most important features among those included in their medical records[1]. In this we build a machine learning classifier to predict a patient's survival. The goal is to select the most important features for predicting the patient's survival. Data for the analysis is available in `heart_failure_clinical_records_dataset.csv`. The data description is to be found in the table `data_description.csv`\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Feature</th>\n",
       "      <th>Explanation</th>\n",
       "      <th>Measurement</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Age</td>\n",
       "      <td>Age of patient</td>\n",
       "      <td>years</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Anaemia</td>\n",
       "      <td>Decrease of red blood cells or hemoglobin</td>\n",
       "      <td>Boolean</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>High blood pressure</td>\n",
       "      <td>If a patient has hypertension</td>\n",
       "      <td>Boolean</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Creatinine phosphokinase</td>\n",
       "      <td>Level of the CPK enzyme in the blood</td>\n",
       "      <td>mcg/L</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Diabetes</td>\n",
       "      <td>If the patient has diabetes</td>\n",
       "      <td>Boolean</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Ejection fraction</td>\n",
       "      <td>Percentage of blood leaving the heart at each ...</td>\n",
       "      <td>Percentage</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Sex</td>\n",
       "      <td>Woman or Man</td>\n",
       "      <td>Binary</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Platelets</td>\n",
       "      <td>Platelets in the blood</td>\n",
       "      <td>kiloplatelets/mL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Serum creatinine</td>\n",
       "      <td>Level of creatinine in the blood</td>\n",
       "      <td>mg/dL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Serum sodium</td>\n",
       "      <td>Level of sodium in the blood</td>\n",
       "      <td>mEq/L</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Smoking</td>\n",
       "      <td>If the patient smokes</td>\n",
       "      <td>Boolean</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>Time</td>\n",
       "      <td>Follow-up period</td>\n",
       "      <td>Days</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>death event</td>\n",
       "      <td>If the patient died during the follow-up period</td>\n",
       "      <td>Boolean</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     Feature  \\\n",
       "0                        Age   \n",
       "1                    Anaemia   \n",
       "2        High blood pressure   \n",
       "3   Creatinine phosphokinase   \n",
       "4                   Diabetes   \n",
       "5          Ejection fraction   \n",
       "6                        Sex   \n",
       "7                  Platelets   \n",
       "8           Serum creatinine   \n",
       "9               Serum sodium   \n",
       "10                   Smoking   \n",
       "11                      Time   \n",
       "12               death event   \n",
       "\n",
       "                                          Explanation       Measurement  \n",
       "0                                      Age of patient             years  \n",
       "1           Decrease of red blood cells or hemoglobin           Boolean  \n",
       "2                       If a patient has hypertension           Boolean  \n",
       "3                Level of the CPK enzyme in the blood             mcg/L  \n",
       "4                         If the patient has diabetes           Boolean  \n",
       "5   Percentage of blood leaving the heart at each ...        Percentage  \n",
       "6                                        Woman or Man            Binary  \n",
       "7                              Platelets in the blood  kiloplatelets/mL  \n",
       "8                    Level of creatinine in the blood             mg/dL  \n",
       "9                        Level of sodium in the blood             mEq/L  \n",
       "10                              If the patient smokes           Boolean  \n",
       "11                                   Follow-up period              Days  \n",
       "12    If the patient died during the follow-up period           Boolean  "
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#given code for data description\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "md = pd.read_csv('data/data_description.csv', sep=';')\n",
    "md"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "### Inspect the data\n",
    " \n",
    "\n",
    "Data is loaded. The `time` feature is dropped, since this is not a biometric.  After loading the data the following questions are answered by the aid of creating meaningfull visualizations and overviews. \n",
    "\n",
    "1. Is the dataset balanced? How might that affect the confusion matrix\n",
    "2. Based on the datatype of each feature, are there any features that cannot be used in a logistic regression classifier? \n",
    "3. Is imputation needed?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>age</th>\n",
       "      <th>anaemia</th>\n",
       "      <th>creatinine_phosphokinase</th>\n",
       "      <th>diabetes</th>\n",
       "      <th>ejection_fraction</th>\n",
       "      <th>high_blood_pressure</th>\n",
       "      <th>platelets</th>\n",
       "      <th>serum_creatinine</th>\n",
       "      <th>serum_sodium</th>\n",
       "      <th>sex</th>\n",
       "      <th>smoking</th>\n",
       "      <th>DEATH_EVENT</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>75.0</td>\n",
       "      <td>0</td>\n",
       "      <td>582</td>\n",
       "      <td>0</td>\n",
       "      <td>20</td>\n",
       "      <td>1</td>\n",
       "      <td>265000.00</td>\n",
       "      <td>1.9</td>\n",
       "      <td>130</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>55.0</td>\n",
       "      <td>0</td>\n",
       "      <td>7861</td>\n",
       "      <td>0</td>\n",
       "      <td>38</td>\n",
       "      <td>0</td>\n",
       "      <td>263358.03</td>\n",
       "      <td>1.1</td>\n",
       "      <td>136</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>65.0</td>\n",
       "      <td>0</td>\n",
       "      <td>146</td>\n",
       "      <td>0</td>\n",
       "      <td>20</td>\n",
       "      <td>0</td>\n",
       "      <td>162000.00</td>\n",
       "      <td>1.3</td>\n",
       "      <td>129</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>50.0</td>\n",
       "      <td>1</td>\n",
       "      <td>111</td>\n",
       "      <td>0</td>\n",
       "      <td>20</td>\n",
       "      <td>0</td>\n",
       "      <td>210000.00</td>\n",
       "      <td>1.9</td>\n",
       "      <td>137</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>65.0</td>\n",
       "      <td>1</td>\n",
       "      <td>160</td>\n",
       "      <td>1</td>\n",
       "      <td>20</td>\n",
       "      <td>0</td>\n",
       "      <td>327000.00</td>\n",
       "      <td>2.7</td>\n",
       "      <td>116</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    age  anaemia  creatinine_phosphokinase  diabetes  ejection_fraction  \\\n",
       "0  75.0        0                       582         0                 20   \n",
       "1  55.0        0                      7861         0                 38   \n",
       "2  65.0        0                       146         0                 20   \n",
       "3  50.0        1                       111         0                 20   \n",
       "4  65.0        1                       160         1                 20   \n",
       "\n",
       "   high_blood_pressure  platelets  serum_creatinine  serum_sodium  sex  \\\n",
       "0                    1  265000.00               1.9           130    1   \n",
       "1                    0  263358.03               1.1           136    1   \n",
       "2                    0  162000.00               1.3           129    1   \n",
       "3                    0  210000.00               1.9           137    1   \n",
       "4                    0  327000.00               2.7           116    0   \n",
       "\n",
       "   smoking  DEATH_EVENT  \n",
       "0        0            1  \n",
       "1        0            1  \n",
       "2        1            1  \n",
       "3        0            1  \n",
       "4        0            1  "
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Load the data. Drop the time feature, since this is not a biometric\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "df = pd.read_csv('data/heart_failure_clinical_records_dataset.csv')\n",
    "df = df.drop(['time'],axis = 1) \n",
    "df.head()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 299 entries, 0 to 298\n",
      "Data columns (total 12 columns):\n",
      " #   Column                    Non-Null Count  Dtype  \n",
      "---  ------                    --------------  -----  \n",
      " 0   age                       299 non-null    float64\n",
      " 1   anaemia                   299 non-null    int64  \n",
      " 2   creatinine_phosphokinase  299 non-null    int64  \n",
      " 3   diabetes                  299 non-null    int64  \n",
      " 4   ejection_fraction         299 non-null    int64  \n",
      " 5   high_blood_pressure       299 non-null    int64  \n",
      " 6   platelets                 299 non-null    float64\n",
      " 7   serum_creatinine          299 non-null    float64\n",
      " 8   serum_sodium              299 non-null    int64  \n",
      " 9   sex                       299 non-null    int64  \n",
      " 10  smoking                   299 non-null    int64  \n",
      " 11  DEATH_EVENT               299 non-null    int64  \n",
      "dtypes: float64(3), int64(9)\n",
      "memory usage: 28.2 KB\n"
     ]
    }
   ],
   "source": [
    "# are there any features that cannot be used in a logistic regression classifier? \n",
    "df.info() \n",
    "# all datatypes needs to be numeric to do the calculations. \n",
    "# All datatypes are float or integers and therefore can be used in a classifier. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of survived:  203\n",
      "Number of deaths:  96\n",
      "\n",
      "\n",
      "% of survived 67.9 %\n",
      "% of death 32.1 %\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df['DEATH_EVENT'].value_counts() \n",
    "# Because of the imbalance of the dataset, \n",
    "# all the methods obtain a better prediction scores on the true negative rate (predicting death), \n",
    "# rather than on the true positive rate\n",
    "\n",
    "survived, dead = df['DEATH_EVENT'].value_counts()\n",
    "print('Number of survived: ', survived)  \n",
    "print('Number of deaths: ', dead) \n",
    "\n",
    "print('\\n')\n",
    "print('% of survived', round(survived / len(df) * 100, 1), '%')\n",
    "print('% of death', round(dead / len(df) * 100, 1), '%')\n",
    "\n",
    "\n",
    "df.DEATH_EVENT.value_counts().plot(kind='barh')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "age                         0\n",
       "anaemia                     0\n",
       "creatinine_phosphokinase    0\n",
       "diabetes                    0\n",
       "ejection_fraction           0\n",
       "high_blood_pressure         0\n",
       "platelets                   0\n",
       "serum_creatinine            0\n",
       "serum_sodium                0\n",
       "sex                         0\n",
       "smoking                     0\n",
       "DEATH_EVENT                 0\n",
       "dtype: int64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# missing data\n",
    "df.isnull().sum() \n",
    "# no missing data, no imputation needed"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature selection\n",
    "\n",
    "We create a meaningful visualization to estimate the three most important features. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "c = df.corr().abs()\n",
    "sns.heatmap(c)\n",
    "# age, ejection_fraction, serum_creatinine do have most light colors -> strongest contributers?\n",
    "# a model with biometrics is for medical use. It is often expensive to acquire the data, \n",
    "# a simpel model which has an acceptable performance metrics is the most practical. Feature \n",
    "# selection might also improve overfitting"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 439.5x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "sns.displot(df, x=\"serum_creatinine\", hue=\"DEATH_EVENT\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Build a classifier\n",
    "\n",
    "\n",
    "We will build a logistic regression classifier. Next we will evaluate the logistic regression classifier using the confusion matrix. Lastly we create a new patient and predict if the patient will survive. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "The df dataframe is the dataframe containing all the raw data (numbers). We will use this to fill the feature matrix $X$ and the vector $y$ with the labeled class. \n",
    "\n",
    "\n",
    "\\begin{equation}\n",
    "X = \n",
    "   \\begin{bmatrix} \\\n",
    "    x_1^{(1)}  & x_2^{(1)} & x_3^{(1)} & .. & x_n^{(1)}\\\\\n",
    "    x_1^{(2)}  & x_2^{(2)} & x_3^{(2)} & .. & x_n^{(2)}\\\\ \n",
    "    x_1^{(3)}  & x_2^{(3)} & x_3^{(3)} & .. & x_n^{(3)} \\\\ \n",
    "    .. & .. & .. & .. & ..\\\\ \n",
    "    x_1^{(m)}  & x_2^{(m)} & x_3^{(m)}  & .. & x_n^{(m)}\\\\ \n",
    "   \\end{bmatrix} \n",
    "   \\\n",
    "   %\n",
    "   y = \n",
    "   \\begin{bmatrix} \\\n",
    "   y^{(1)} \\\\\n",
    "   y^{(2)} \\\\ \n",
    "   y^{(3)} \\\\ \n",
    "   .. \\\\ \n",
    "   y^{(m)} \\\\ \n",
    "   \\end{bmatrix} \n",
    "  %\n",
    "\\end{equation}\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LogisticRegression()"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Build model \n",
    "#preprosses data \n",
    "y = np.array(df['DEATH_EVENT'])\n",
    "X = df.iloc[:,[0,1,2,3,4,5,6,7,8,9,10]] #without feature selection\n",
    "\n",
    "#normalise \n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "def normalize(X):\n",
    "    scalar = StandardScaler()\n",
    "    scalar = scalar.fit(X)\n",
    "    X = scalar.transform(X)\n",
    "    return X\n",
    "\n",
    "X = normalize(X)\n",
    "X.shape\n",
    "\n",
    "#split \n",
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2)\n",
    "\n",
    "#train  \n",
    "from sklearn.linear_model import LogisticRegression\n",
    "model = LogisticRegression()\n",
    "model.fit(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.6021  age\n",
      "0.3509  anaemia\n",
      "0.2897  creatinine_phosphokinase\n",
      "0.02878  diabetes\n",
      "-1.04  ejection_fraction\n",
      "0.2561  high_blood_pressure\n",
      "-0.155  platelets\n",
      "0.8209  serum_creatinine\n",
      "-0.187  serum_sodium\n",
      "-0.1607  sex\n",
      "-0.007137  smoking\n"
     ]
    }
   ],
   "source": [
    "# optional verify feature weights\n",
    "for i in model.coef_:\n",
    "       for index, j in enumerate(i):        \n",
    "            print(f\"{j:.4}  {list(df.columns.values)[index]}\")\n",
    "# You will see that age, ejection_fraction and serum_creatinine are driving the outcome the most"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[35  5]\n",
      " [11  9]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.76      0.88      0.81        40\n",
      "           1       0.64      0.45      0.53        20\n",
      "\n",
      "    accuracy                           0.73        60\n",
      "   macro avg       0.70      0.66      0.67        60\n",
      "weighted avg       0.72      0.73      0.72        60\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# evaluate the model\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "y_pred = model.predict(X_test)\n",
    "confusion_matrix = confusion_matrix(y_test, y_pred)\n",
    "print(confusion_matrix)\n",
    "print(classification_report(y_test, y_pred))\n",
    "# Accuracy is not that high Indeed negatives are predicted better then positives. \n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "test  acc: 0.7333333333333333\n",
      "train acc: 0.7866108786610879\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    " # function to evaluate\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from sklearn.metrics import mean_squared_error\n",
    "from sklearn.model_selection import learning_curve\n",
    "\n",
    "def plot_learning_curves(model, X_train, y_train, X_val, y_val):\n",
    "    \"\"\"\n",
    "    input:\n",
    "        model:pipeline object\n",
    "        X_train, y_train: training data\n",
    "        X_val, y_val: test data\n",
    "    \"\"\"\n",
    "    train_errors, val_errors = [], []\n",
    "    for m in range(30, len(X_train)):\n",
    "        model.fit(X_train[:m], y_train[:m])\n",
    "        y_train_predict = model.predict(X_train[:m])\n",
    "        y_val_predict = model.predict(X_val)\n",
    "        train_errors.append(mean_squared_error(y_train_predict, y_train[:m]))\n",
    "        val_errors.append(mean_squared_error(y_val_predict, y_val))\n",
    "\n",
    "    plt.plot(np.sqrt(train_errors), \"r-+\", linewidth=1, label=\"training data\")\n",
    "    plt.plot(np.sqrt(val_errors), \"b-\", linewidth=1, label=\"validation data\")\n",
    "    plt.legend(loc=\"upper right\", fontsize=10)   \n",
    "    plt.xlabel(\"Training set size\", fontsize=10) \n",
    "    plt.ylabel(\"RMSE\", fontsize=10)     \n",
    "    # compare accuracy train versus test to access overfit \n",
    "    print(f'test  acc: {model.score(X_val, y_val)}')\n",
    "    print(f'train acc: {model.score(X_train, y_train)}')\n",
    "    \n",
    "    \n",
    "plot_learning_curves(model, X_train, y_train, X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[39  4]\n",
      " [ 8  9]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.83      0.91      0.87        43\n",
      "           1       0.69      0.53      0.60        17\n",
      "\n",
      "    accuracy                           0.80        60\n",
      "   macro avg       0.76      0.72      0.73        60\n",
      "weighted avg       0.79      0.80      0.79        60\n",
      "\n",
      "[[ 0.59047315 -0.89701892  0.70281006]]\n"
     ]
    }
   ],
   "source": [
    "# and or new model with feature selection \n",
    "y = np.array(df['DEATH_EVENT'])\n",
    "X = df.iloc[:,[0,4,7]]\n",
    "#normalise\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "def normalize(X):\n",
    "    scalar = StandardScaler()\n",
    "    scalar = scalar.fit(X)\n",
    "    X = scalar.transform(X)\n",
    "    return X\n",
    "\n",
    "X = normalize(X)\n",
    "#split\n",
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2)\n",
    "\n",
    "#train LogisticRegression\n",
    "new_model = LogisticRegression()\n",
    "new_model.fit(X_train, y_train)\n",
    "# evaluate the model. Is your assumption of contribution features correct based on the outcome?\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "y_pred = new_model.predict(X_test)\n",
    "confusion_matrix = confusion_matrix(y_test, y_pred)\n",
    "print(confusion_matrix)\n",
    "print(classification_report(y_test, y_pred))\n",
    "# optional\n",
    "\n",
    "print(new_model.coef_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "survivalrate: 0.999\n"
     ]
    }
   ],
   "source": [
    "print(f\"survivalrate: {round(new_model.predict_proba([[25, 25, 2.0]])[0][0], 3)}\") "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### improving the model\n",
    "\n",
    "What might be another classifier suitable for the dataset and the problem? "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "#1 Decision trees often perform well on imbalanced datasets because their hierarchical structure allows \n",
    "#  them to learn signals from both classes.\n",
    "#2 cross validation is advisable because of the small sample. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier, BaggingClassifier, AdaBoostClassifier, VotingClassifier"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[40  3]\n",
      " [ 8  9]]\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.83      0.93      0.88        43\n",
      "           1       0.75      0.53      0.62        17\n",
      "\n",
      "    accuracy                           0.82        60\n",
      "   macro avg       0.79      0.73      0.75        60\n",
      "weighted avg       0.81      0.82      0.81        60\n",
      "\n"
     ]
    }
   ],
   "source": [
    "# improve model for instance with AdaBoost (applicabl for binary classes)\n",
    "adb = AdaBoostClassifier(LogisticRegression(), n_estimators = 10, learning_rate = 1)\n",
    "adb.fit(X_train, y_train)\n",
    "\n",
    "y_pred = adb.predict(X_test)\n",
    "\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.metrics import confusion_matrix\n",
    "confusion_matrix = confusion_matrix(y_test, y_pred)\n",
    "print(confusion_matrix)\n",
    "print(classification_report(y_test, y_pred))\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}