titanic.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "407b00b8",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "0226df5b",
   "metadata": {},
   "outputs": [],
   "source": [
    "titanic_data = pd.read_csv('data/train.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a1169e2d",
   "metadata": {},
   "outputs": [],
   "source": [
    "import seaborn as sns\n",
    "\n",
    "sns.heatmap(titanic_data.corr(), cmap=\"YlGnBu\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "7584cf77",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import StratifiedShuffleSplit\n",
    "\n",
    "split = StratifiedShuffleSplit(n_splits=1, test_size=0.2)\n",
    "for train_indices, test_indices in split.split(titanic_data, titanic_data[[\"Survived\", \"Pclass\", \"Sex\"]]):\n",
    "    strat_train_set = titanic_data.loc[train_indices]\n",
    "    strat_test_set = titanic_data.loc[test_indices]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "01d3b125",
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.subplot(1,2,1)\n",
    "strat_train_set['Survived'].hist()\n",
    "strat_train_set['Pclass'].hist()\n",
    "\n",
    "plt.subplot(1,2,2)\n",
    "strat_test_set['Survived'].hist()\n",
    "strat_test_set['Pclass'].hist()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "adf1d10a",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.base import BaseEstimator, TransformerMixin\n",
    "from sklearn.impute import SimpleImputer\n",
    "\n",
    "class AgeImputer(BaseEstimator, TransformerMixin):\n",
    "    \n",
    "    def fit(self, X, y=None):\n",
    "        return self\n",
    "    \n",
    "    def transform(self, X):\n",
    "        imputer = SimpleImputer(strategy=\"mean\")\n",
    "        X[\"Age\"] = imputer.fit_transform(X[['Age']])\n",
    "        return X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "728ab987",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import OneHotEncoder\n",
    "\n",
    "class FeatureEncoder(BaseEstimator, TransformerMixin):\n",
    "    \n",
    "    def fit(self, X, y=None):\n",
    "        return self\n",
    "    \n",
    "    def transform(self, X):\n",
    "        encoder = OneHotEncoder()\n",
    "        matrix = encoder.fit_transform(X[[\"Embarked\"]]).toarray()\n",
    "        \n",
    "        column_names = [\"C\", \"S\", \"Q\", \"N\"]\n",
    "        \n",
    "        for i in range(len(matrix.T)):\n",
    "            X[column_names[i]] = matrix.T[i]\n",
    "            \n",
    "        matrix = encoder.fit_transform(X[['Sex']]).toarray()   \n",
    "        \n",
    "        column_names = ['Female', 'Male']\n",
    "        \n",
    "        for i in range(len(matrix.T)):\n",
    "            X[column_names[i]] = matrix.T[i]\n",
    "            \n",
    "        return X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "ba584c0f",
   "metadata": {},
   "outputs": [],
   "source": [
    "class FeatureDropper(BaseEstimator, TransformerMixin):\n",
    "    \n",
    "    def fit(self, X, y=None):\n",
    "        return self\n",
    "    \n",
    "    def transform(self, X):\n",
    "        return X.drop([\"Embarked\", \"Name\", \"Ticket\", \"Cabin\", \"Sex\", \"N\"], axis=1, errors=\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "f2557f16",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.pipeline import Pipeline\n",
    "\n",
    "pipeline = Pipeline([(\"ageimputer\", AgeImputer()),\n",
    "                     (\"featureencoder\", FeatureEncoder()),\n",
    "                     (\"featuredropper\", FeatureDropper())])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "f06b604f",
   "metadata": {},
   "outputs": [],
   "source": [
    "strat_train_set = pipeline.fit_transform(strat_train_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "c82b3476",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import StandardScaler\n",
    "\n",
    "X = strat_train_set.drop(['Survived'], axis=1)\n",
    "y = strat_train_set['Survived']\n",
    "\n",
    "scaler = StandardScaler()\n",
    "X_data = scaler.fit_transform(X)\n",
    "y_data = y.to_numpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "69c4e0e7",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "\n",
    "clf = RandomForestClassifier()\n",
    "\n",
    "param_grid = [\n",
    "    {\"n_estimators\": [10, 100, 200, 500],\n",
    "     \"max_depth\": [None, 5, 10], \n",
    "     \"min_samples_split\": [2, 3, 4]\n",
    "    }\n",
    "]\n",
    "\n",
    "grid_search = GridSearchCV(clf, param_grid, cv=3, scoring=\"accuracy\", return_train_score=True)\n",
    "grid_search.fit(X_data, y_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "1620ae4a",
   "metadata": {},
   "outputs": [],
   "source": [
    "final_clf = grid_search.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "635aa5d9",
   "metadata": {},
   "outputs": [],
   "source": [
    "strat_test_set = pipeline.fit_transform(strat_test_set)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "ac345fd3",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_test = strat_test_set.drop(['Survived'], axis=1)\n",
    "y_test = strat_test_set[\"Survived\"]\n",
    "\n",
    "scaler = StandardScaler()\n",
    "X_data_test = scaler.fit_transform(X_test)\n",
    "y_data_test = y_test.to_numpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fde31d12",
   "metadata": {},
   "outputs": [],
   "source": [
    "final_clf.score(X_data_test, y_data_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "56c7203f",
   "metadata": {},
   "outputs": [],
   "source": [
    "final_data = pipeline.fit_transform(titanic_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "533e61a3",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_final = final_data.drop(['Survived'], axis=1)\n",
    "y_final = final_data[\"Survived\"]\n",
    "\n",
    "scaler = StandardScaler()\n",
    "X_data_final = scaler.fit_transform(X_final)\n",
    "y_data_final = y_final.to_numpy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6f8696ff",
   "metadata": {},
   "outputs": [],
   "source": [
    "prod_clf = RandomForestClassifier()\n",
    "\n",
    "param_grid = [\n",
    "    {\"n_estimators\": [10, 100, 200, 500],\n",
    "     \"max_depth\": [None, 5, 10], \n",
    "     \"min_samples_split\": [2, 3, 4]\n",
    "    }\n",
    "]\n",
    "\n",
    "grid_search = GridSearchCV(prod_clf, param_grid, cv=3, scoring=\"accuracy\", return_train_score=True)\n",
    "grid_search.fit(X_data_final, y_data_final)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2bfb6561",
   "metadata": {},
   "outputs": [],
   "source": [
    "prod_final_clf = grid_search.best_estimator_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "15e3cb5a",
   "metadata": {},
   "outputs": [],
   "source": [
    "titanic_test_data = pd.read_csv(\"data/test.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "id": "8b1323dc",
   "metadata": {},
   "outputs": [],
   "source": [
    "final_test_data = pipeline.fit_transform(titanic_test_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "cb5773e8",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_final_test = final_test_data.fillna(method=\"ffill\")\n",
    "scaler = StandardScaler()\n",
    "X_data_final_test = scaler.fit_transform(X_final_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "0a6fa0e0",
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = prod_final_clf.predict(X_data_final_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "id": "3844039f",
   "metadata": {},
   "outputs": [],
   "source": [
    "final_df = pd.DataFrame(titanic_test_data['PassengerId'])\n",
    "final_df['Survived'] = predictions\n",
    "final_df.to_csv(\"data/predictions.csv\", index=False)"
   ]
  }
 ],
 "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": "3.10.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}