diff --git a/logistic_regression.ipynb b/logistic_regression.ipynb index c76ab2c..ea4b8a1 100644 --- a/logistic_regression.ipynb +++ b/logistic_regression.ipynb @@ -1627,6 +1627,241 @@ "En pratique, les deux modèles peuvent être retenus selon le compromis souhaité entre performance brute et simplicité structurelle. \n", "Le modèle L1 est particulièrement pertinent dans un objectif de généralisation ou de sélection automatique de variables.\n" ] + }, + { + "cell_type": "markdown", + "id": "d11139d7", + "metadata": {}, + "source": [ + "# Étape 13 (bonus): application de l'ACP pour essayer d'améliorer les modèles" + ] + }, + { + "cell_type": "code", + "execution_count": 61, + "id": "b3a8341b", + "metadata": {}, + "outputs": [ + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Meilleurs hyperparamètres : {'C': 1.6681005372000592, 'penalty': 'l2', 'solver': 'liblinear'}\n", + "F1-score moyen (validation croisée) : 0.715\n", + "\n", + " Évaluation sur le jeu de test :\n", + "Accuracy : 0.833\n", + "F1-score : 0.818\n", + "AUC : 0.832\n", + "\n", + "Classification Report :\n", + " precision recall f1-score support\n", + "\n", + " 0 0.85 0.85 0.85 13\n", + " 1 0.82 0.82 0.82 11\n", + "\n", + " accuracy 0.83 24\n", + " macro avg 0.83 0.83 0.83 24\n", + "weighted avg 0.83 0.83 0.83 24\n", + "\n" + ] + }, + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "data": { + "image/png": "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", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "from sklearn.decomposition import PCA\n", + "\n", + "\n", + "# Étape 1 — Sélection du meilleur nombre de composantes via validation croisée\n", + "scores = []\n", + "components_range = range(2, X_train_scaled.shape[1] + 1)\n", + "\n", + "for n in components_range:\n", + " pca = PCA(n_components=n, svd_solver='full', random_state=42)\n", + " X_train_pca = pca.fit_transform(X_train_scaled)\n", + " model = LogisticRegression(random_state=42, max_iter=1000)\n", + " score = cross_val_score(model, X_train_pca, y_train, cv=10, scoring='f1').mean()\n", + " scores.append(score)\n", + "\n", + "# Tracé des performances\n", + "plt.figure(figsize=(8, 5))\n", + "plt.plot(components_range, scores, marker='o')\n", + "plt.title(\"Sélection du nombre de composantes pour PCA\")\n", + "plt.xlabel(\"Nombre de composantes principales\")\n", + "plt.ylabel(\"F1-score (validation croisée)\")\n", + "plt.grid(True)\n", + "plt.tight_layout()\n", + "plt.show()\n", + "\n", + "# Étape 2 — Application du PCA avec le meilleur n\n", + "n_best = components_range[np.argmax(scores)]\n", + "pca_final = PCA(n_components=n_best, svd_solver='full', random_state=42)\n", + "X_train_pca = pca_final.fit_transform(X_train_scaled)\n", + "X_test_pca = pca_final.transform(X_test_scaled)\n", + "\n", + "# Étape 3 — Optimisation du modèle via GridSearchCV\n", + "param_grid = {\n", + " 'penalty': ['l1', 'l2'],\n", + " 'C': np.logspace(-2, 3, 10),\n", + " 'solver': ['liblinear'] # nécessaire pour l1\n", + "}\n", + "logreg = LogisticRegression(random_state=42, max_iter=1000)\n", + "grid_search = GridSearchCV(logreg, param_grid, cv=10, scoring='f1')\n", + "grid_search.fit(X_train_pca, y_train)\n", + "\n", + "best_model = grid_search.best_estimator_\n", + "print(\"Meilleurs hyperparamètres :\", grid_search.best_params_)\n", + "print(\"F1-score moyen (validation croisée) :\", round(grid_search.best_score_, 3))\n", + "\n", + "# Étape 4 — Évaluation finale sur le jeu de test\n", + "y_pred_pca = best_model.predict(X_test_pca)\n", + "y_proba_pca = best_model.predict_proba(X_test_pca)[:, 1]\n", + "\n", + "print(\"\\n Évaluation sur le jeu de test :\")\n", + "print(\"Accuracy :\", round(accuracy_score(y_test, y_pred_pca), 3))\n", + "print(\"F1-score :\", round(f1_score(y_test, y_pred_pca), 3))\n", + "print(\"AUC :\", round(roc_auc_score(y_test, y_proba_pca), 3))\n", + "print(\"\\nClassification Report :\\n\", classification_report(y_test, y_pred_pca))\n", + "\n", + "# Matrice de confusion\n", + "cm = confusion_matrix(y_test, y_pred_pca)\n", + "plt.figure(figsize=(6, 5))\n", + "sns.heatmap(cm, annot=True, fmt='d', cmap='Blues', cbar=False,\n", + " xticklabels=['Sain', 'Malade'], yticklabels=['Sain', 'Malade'])\n", + "plt.xlabel(\"Prédictions\")\n", + "plt.ylabel(\"Réel\")\n", + "plt.title(\"Matrice de confusion (PCA)\")\n", + "plt.tight_layout()\n", + "plt.show()\n", + "\n", + "# Courbe ROC\n", + "fpr, tpr, _ = roc_curve(y_test, y_proba_pca)\n", + "auc_value = roc_auc_score(y_test, y_proba_pca)\n", + "\n", + "plt.figure(figsize=(8, 6))\n", + "plt.plot(fpr, tpr, label=f\"AUC = {auc_value:.2f}\", color='blue')\n", + "plt.plot([0, 1], [0, 1], linestyle='--', color='red')\n", + "plt.xlabel(\"Taux de faux positifs (FPR)\")\n", + "plt.ylabel(\"Taux de vrais positifs (TPR)\")\n", + "plt.title(\"Courbe ROC — Modèle après PCA\")\n", + "plt.grid(True)\n", + "plt.legend()\n", + "plt.tight_layout()\n", + "plt.show()\n" + ] + }, + { + "cell_type": "markdown", + "id": "394a2a73", + "metadata": {}, + "source": [ + "\n", + "L'objectif de cette étape est d'examiner si une réduction de dimension par Analyse en Composantes Principales (PCA) peut améliorer les performances du modèle de classification.\n", + "\n", + "---\n", + "\n", + "### Justification\n", + "\n", + "Même après transformation logarithmique et standardisation, les variables explicatives peuvent contenir des corrélations redondantes ou du bruit peu informatif. Le PCA permet de :\n", + "- projeter les données dans un nouvel espace de dimension réduite,\n", + "- concentrer l’essentiel de la variance sur quelques composantes principales,\n", + "- stabiliser l’apprentissage sur petits jeux de données en atténuant la variance inutile.\n", + "\n", + "---\n", + "\n", + "### Méthodologie\n", + "\n", + "1. Le PCA est appliqué sur les données standardisées (`X_train_scaled`) via `sklearn.decomposition.PCA`.\n", + "2. Les performances de classification sont évaluées en fonction du nombre de composantes principales, de 2 à 9 (limite fixée par le nombre de variables d'origine).\n", + "3. À chaque itération, un modèle de régression logistique est entraîné avec validation croisée (cv=10) pour mesurer le F1-score moyen.\n", + "4. Le nombre optimal de composantes est sélectionné en maximisant ce score.\n", + "5. Enfin, le modèle optimal est réentraîné sur l’ensemble réduit et évalué sur le jeu de test.\n", + "\n", + "---\n", + "\n", + "### Sélection du nombre de composantes\n", + "\n", + "La figure ci-dessous montre l'évolution du F1-score moyen en fonction du nombre de composantes principales :\n", + "\n", + "> *Figure – Sélection du nombre de composantes (PCA)*\n", + "\n", + "Le maximum est atteint à 7 composantes, avec un F1-score de validation croisée de 0.715. Au-delà, le score stagne, ce qui indique que les dernières composantes contiennent peu d'information discriminante utile.\n", + "\n", + "---\n", + "\n", + "### Résultats finaux sur le jeu de test\n", + "\n", + "Après réduction à 7 composantes, puis optimisation des hyperparamètres de la régression logistique (`C`, `penalty`) via `GridSearchCV`, les résultats obtenus sont les suivants :\n", + "\n", + "- Accuracy : 0.833\n", + "- F1-score : 0.818\n", + "- AUC : 0.832\n", + "\n", + "#### Classification report :\n", + "\n", + "| Classe | Précision | Rappel | F1-score | Support |\n", + "|--------------|-----------|--------|----------|---------|\n", + "| 0 (sain) | 0.85 | 0.85 | 0.85 | 13 |\n", + "| 1 (malade) | 0.82 | 0.82 | 0.82 | 11 |\n", + "| Macro avg | 0.83 | 0.83 | 0.83 | 24 |\n", + "\n", + "---\n", + "\n", + "### Matrice de confusion\n", + "\n", + "| | Prédit : sain | Prédit : malade |\n", + "|---------------|----------------|------------------|\n", + "| Réel : sain | 11 | 2 |\n", + "| Réel : malade | 2 | 9 |\n", + "\n", + "Le modèle après PCA parvient à bien équilibrer les deux classes, avec un nombre réduit de faux positifs et de faux négatifs.\n", + "\n", + "---\n", + "\n", + "### Courbe ROC\n", + "\n", + "La courbe ROC confirme une capacité discriminante élevée, avec une AUC de 0.83. Cela signifie que le modèle est capable, dans 83 % des cas, d’attribuer une probabilité plus élevée à une observation positive qu’à une observation négative.\n", + "\n", + "---\n", + "\n", + "### Conclusion\n", + "\n", + "L’application du PCA classique avec 7 composantes principales :\n", + "- améliore légèrement les performances globales par rapport aux modèles sans réduction de dimension,\n", + "- stabilise l’apprentissage sur un jeu de données restreint,\n", + "- conserve une bonne capacité discriminante tout en simplifiant la représentation des données.\n", + "\n", + "Ces résultats valident l’intérêt de la réduction de dimension par PCA linéaire, et justifient son intégration finale dans le pipeline supervisé.\n" + ] } ], "metadata": {