Ансамбли.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "3085ec43-34bb-4afb-9515-517e0489db3e",
   "metadata": {},
   "source": [
    "## Ансамбли и полносвязные нейронные сети\n",
    "В этом ноутбуке вам нужно обучить модели на датасете классификации из предыдущего ноутбука и сравнить результаты. Вам будет предоставлен baseline, на основе которого вы будете доделывать предсказывающие модели. Оценка лабы будет зависеть от ROC-AUC на тестовых данных по следующим критериям:\n",
    "\\\n",
    "AUC - на тестовых данных\n",
    "- $AUC \\leq 0.76$ - 0 баллов\n",
    "- $0.76 < AUC \\leq 0.77$ - 2 балла\n",
    "- $0.77 < AUC \\leq 0.78$ - 4 балла\n",
    "- $0.78 < AUC \\leq 0.79$ - 6 баллов\n",
    "- $0.79 < AUC \\leq 0.80$ - 8 баллов\n",
    "- $AUC > 0.80$ - 10 баллов\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ec07e3a2-480a-4350-868e-02679ff2aada",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier\n",
    "from sklearn.neural_network import MLPClassifier\n",
    "\n",
    "from sklearn.metrics import roc_auc_score, precision_score, recall_score, roc_curve, accuracy_score\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "563ad31b-5c83-4366-819a-34dad4edecdc",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "data = pd.read_csv('german.csv', sep=';')\n",
    "print(data.head())\n",
    "\n",
    "X = data.iloc[:, 1:].to_numpy()\n",
    "y = data.iloc[:, 0].to_numpy()\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f93737ec-e5eb-4d72-8beb-5dba4d4c581f",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "plt.hist(y_train, bins=2, edgecolor='k')\n",
    "plt.xticks([0, 1])\n",
    "plt.xlabel('Class (0: Non-Creditworthy, 1: Creditworthy)')\n",
    "plt.ylabel('Count')\n",
    "plt.title('Distribution of Classes in Training Data')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c1cf8d00-92a3-4b62-bca4-d854b72574d1",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Обучение Random Forest\n",
    "rf_model = RandomForestClassifier(n_estimators=100, random_state=42)\n",
    "rf_model.fit(X_train, y_train)\n",
    "\n",
    "# Прогноз на тестовых данных\n",
    "rf_pred = rf_model.predict(X_test)\n",
    "\n",
    "# Расчет метрик для Random Forest\n",
    "rf_roc_auc = roc_auc_score(y_test, rf_pred)\n",
    "rf_accuracy = accuracy_score(y_test, rf_pred)\n",
    "rf_precision = precision_score(y_test, rf_pred)\n",
    "rf_recall = recall_score(y_test, rf_pred)\n",
    "\n",
    "print(\"Random Forest метрики:\")\n",
    "print(f\"ROC AUC: {rf_roc_auc:.2f}\")\n",
    "print(f\"Accuracy: {rf_accuracy:.2f}\")\n",
    "print(f\"Precision: {rf_precision:.2f}\")\n",
    "print(f\"Recall: {rf_recall:.2f}\")\n",
    "\n",
    "# Обучение Gradient Boosting\n",
    "gb_model = GradientBoostingClassifier(n_estimators=100, random_state=42)\n",
    "gb_model.fit(X_train, y_train)\n",
    "\n",
    "# Прогноз на тестовых данных\n",
    "gb_pred = gb_model.predict(X_test)\n",
    "\n",
    "# Расчет метрик для Gradient Boosting\n",
    "gb_roc_auc = roc_auc_score(y_test, gb_pred)\n",
    "gb_accuracy = accuracy_score(y_test, gb_pred)\n",
    "gb_precision = precision_score(y_test, gb_pred)\n",
    "gb_recall = recall_score(y_test, gb_pred)\n",
    "\n",
    "print(\"\\nGradient Boosting метрики:\")\n",
    "print(f\"ROC AUC: {gb_roc_auc:.2f}\")\n",
    "print(f\"Accuracy: {gb_accuracy:.2f}\")\n",
    "print(f\"Precision: {gb_precision:.2f}\")\n",
    "print(f\"Recall: {gb_recall:.2f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ae1e040c-ddd0-4952-9dcb-58c1226da40a",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# Обучение MLP (Multi-Layer Perceptron) нейронной сети\n",
    "mlp_model = MLPClassifier(hidden_layer_sizes=(30,), max_iter=500, random_state=42)\n",
    "mlp_model.fit(X_train, y_train)\n",
    "\n",
    "# Прогноз на тестовых данных\n",
    "mlp_pred = mlp_model.predict(X_test)\n",
    "\n",
    "# Расчет метрик для MLP нейронной сети\n",
    "mlp_roc_auc = roc_auc_score(y_test, mlp_pred)\n",
    "mlp_accuracy = accuracy_score(y_test, mlp_pred)\n",
    "mlp_precision = precision_score(y_test, mlp_pred)\n",
    "mlp_recall = recall_score(y_test, mlp_pred)\n",
    "\n",
    "print(\"\\nMLP (Neural Network) метрики:\")\n",
    "print(f\"ROC AUC: {mlp_roc_auc:.2f}\")\n",
    "print(f\"Accuracy: {mlp_accuracy:.2f}\")\n",
    "print(f\"Precision: {mlp_precision:.2f}\")\n",
    "print(f\"Recall: {mlp_recall:.2f}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d2d6eb00-77fd-40dc-a3c5-35c1fe0200c0",
   "metadata": {},
   "source": [
    "## Экспериментируйте\n",
    "Для получения лучшего качества придется поэкспериментировать. Подсказка: попробуйте оптимизировать гиперпараметры модели"
   ]
  }
 ],
 "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": "3.11.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}