""" SFM_ch10_scoring ================ Credit Scoring with Logistic Regression Description: - Generate synthetic credit scoring dataset - Fit logistic regression model - Evaluate with ROC curve and AUC - Confusion matrix analysis - Coefficient importance and odds ratios Statistics of Financial Markets course """ import numpy as np import pandas as pd import matplotlib.pyplot as plt from scipy import stats from sklearn.linear_model import LogisticRegression from sklearn.model_selection import train_test_split from sklearn.preprocessing import StandardScaler from sklearn.metrics import (roc_curve, auc, confusion_matrix, classification_report, roc_auc_score) import warnings warnings.filterwarnings('ignore') # --- Standard chart style --- plt.rcParams['figure.facecolor'] = 'none' plt.rcParams['axes.facecolor'] = 'none' plt.rcParams['savefig.facecolor'] = 'none' plt.rcParams['savefig.transparent'] = True plt.rcParams['axes.grid'] = False plt.rcParams['font.family'] = 'sans-serif' plt.rcParams['font.sans-serif'] = ['Helvetica', 'Arial', 'DejaVu Sans'] plt.rcParams['font.size'] = 8 plt.rcParams['axes.labelsize'] = 9 plt.rcParams['axes.titlesize'] = 10 plt.rcParams['axes.spines.top'] = False plt.rcParams['axes.spines.right'] = False plt.rcParams['axes.linewidth'] = 0.5 plt.rcParams['lines.linewidth'] = 0.75 plt.rcParams['legend.facecolor'] = 'none' plt.rcParams['legend.framealpha'] = 0 def save_fig(name): plt.savefig(f'../../charts/{name}.pdf', bbox_inches='tight', transparent=True) plt.savefig(f'../../charts/{name}.png', bbox_inches='tight', transparent=True, dpi=300) plt.close() print(f" Saved: {name}.pdf") print("=" * 70) print("SFM CHAPTER 10: CREDIT SCORING") print("=" * 70) np.random.seed(42) # ============================================================================= # 1. Generate Synthetic Dataset # ============================================================================= print("\n1. GENERATING SYNTHETIC DATASET") print("-" * 40) n = 5000 # Features income = np.random.lognormal(mean=10.5, sigma=0.5, size=n) age = np.random.normal(40, 12, n).clip(18, 75) debt_ratio = np.random.beta(2, 5, n) credit_history = np.random.randint(0, 30, n) num_accounts = np.random.poisson(3, n) employment_years = np.random.exponential(5, n).clip(0, 40) # Default probability (logistic model) z = (-3.0 - 0.00002 * income + 0.02 * age + 3.0 * debt_ratio - 0.1 * credit_history + 0.15 * num_accounts - 0.08 * employment_years + np.random.normal(0, 0.5, n)) prob_default = 1 / (1 + np.exp(-z)) default = (np.random.uniform(0, 1, n) < prob_default).astype(int) df = pd.DataFrame({ 'income': income, 'age': age, 'debt_ratio': debt_ratio, 'credit_history': credit_history, 'num_accounts': num_accounts, 'employment_years': employment_years, 'default': default }) print(f" Total observations: {n}") print(f" Default rate: {default.mean():.4f} ({default.sum()} defaults)") print(f" Features: {list(df.columns[:-1])}") # ============================================================================= # 2. Train/Test Split and Scaling # ============================================================================= print("\n2. PREPARING DATA") print("-" * 40) features = ['income', 'age', 'debt_ratio', 'credit_history', 'num_accounts', 'employment_years'] X = df[features].values y = df['default'].values X_train, X_test, y_train, y_test = train_test_split( X, y, test_size=0.3, random_state=42, stratify=y) scaler = StandardScaler() X_train_sc = scaler.fit_transform(X_train) X_test_sc = scaler.transform(X_test) print(f" Training set: {len(X_train)} " f"({y_train.mean():.4f} default rate)") print(f" Test set: {len(X_test)} " f"({y_test.mean():.4f} default rate)") # ============================================================================= # 3. Logistic Regression # ============================================================================= print("\n3. LOGISTIC REGRESSION") print("-" * 40) model = LogisticRegression(max_iter=1000, random_state=42) model.fit(X_train_sc, y_train) # Predictions y_pred = model.predict(X_test_sc) y_prob = model.predict_proba(X_test_sc)[:, 1] # Coefficients print(f"\n Coefficients (standardized):") print(f" {'Feature':<20} {'Coeff':>10} {'Odds Ratio':>12}") print(" " + "-" * 44) for feat, coef in zip(features, model.coef_[0]): print(f" {feat:<20} {coef:>10.4f} {np.exp(coef):>12.4f}") print(f" {'Intercept':<20} {model.intercept_[0]:>10.4f}") # ============================================================================= # 4. Model Evaluation # ============================================================================= print("\n4. MODEL EVALUATION") print("-" * 40) auc_score = roc_auc_score(y_test, y_prob) print(f" AUC-ROC: {auc_score:.4f}") cm = confusion_matrix(y_test, y_pred) tn, fp, fn, tp = cm.ravel() print(f"\n Confusion Matrix:") print(f" {'':>12} {'Pred No':>10} {'Pred Yes':>10}") print(f" {'Actual No':<12} {tn:>10d} {fp:>10d}") print(f" {'Actual Yes':<12} {fn:>10d} {tp:>10d}") accuracy = (tp + tn) / (tp + tn + fp + fn) precision = tp / (tp + fp) if (tp + fp) > 0 else 0 recall = tp / (tp + fn) if (tp + fn) > 0 else 0 f1 = (2 * precision * recall / (precision + recall) if (precision + recall) > 0 else 0) print(f"\n Accuracy: {accuracy:.4f}") print(f" Precision: {precision:.4f}") print(f" Recall: {recall:.4f}") print(f" F1-Score: {f1:.4f}") # ============================================================================= # 5. FIGURE: Credit Scoring Analysis (4-panel) # ============================================================================= print("\n5. CREATING FIGURE") print("-" * 40) fig, axes = plt.subplots(2, 2, figsize=(14, 10)) # Panel A: ROC Curve fpr, tpr, thresholds = roc_curve(y_test, y_prob) axes[0, 0].plot(fpr, tpr, color='#1A3A6E', linewidth=1.5, label=f'Logistic Regression (AUC = {auc_score:.3f})') axes[0, 0].plot([0, 1], [0, 1], 'k--', linewidth=0.8, label='Random Classifier') axes[0, 0].fill_between(fpr, tpr, alpha=0.1, color='#1A3A6E') axes[0, 0].set_title('ROC Curve', fontweight='bold') axes[0, 0].set_xlabel('False Positive Rate') axes[0, 0].set_ylabel('True Positive Rate') axes[0, 0].legend(loc='lower right', fontsize=8) # Panel B: Confusion Matrix Heatmap im = axes[0, 1].imshow(cm, cmap='Blues', aspect='auto') axes[0, 1].set_xticks([0, 1]) axes[0, 1].set_yticks([0, 1]) axes[0, 1].set_xticklabels(['No Default', 'Default']) axes[0, 1].set_yticklabels(['No Default', 'Default']) axes[0, 1].set_xlabel('Predicted') axes[0, 1].set_ylabel('Actual') axes[0, 1].set_title('Confusion Matrix', fontweight='bold') for i in range(2): for j in range(2): axes[0, 1].text(j, i, f'{cm[i, j]}', ha='center', va='center', fontsize=14, color='white' if cm[i, j] > cm.max() / 2 else 'black') plt.colorbar(im, ax=axes[0, 1], shrink=0.8) # Panel C: Coefficient Importance (horizontal bar) coefs = model.coef_[0] sorted_idx = np.argsort(np.abs(coefs)) axes[1, 0].barh(range(len(features)), coefs[sorted_idx], color=['#DC3545' if c > 0 else '#2E7D32' for c in coefs[sorted_idx]], alpha=0.7) axes[1, 0].set_yticks(range(len(features))) axes[1, 0].set_yticklabels([features[i] for i in sorted_idx]) axes[1, 0].axvline(x=0, color='gray', linewidth=0.5) axes[1, 0].set_title('Feature Coefficients (Standardized)', fontweight='bold') axes[1, 0].set_xlabel('Coefficient') # Panel D: Predicted probability distribution axes[1, 1].hist(y_prob[y_test == 0], bins=50, density=True, alpha=0.5, color='#2E7D32', edgecolor='white', label='No Default') axes[1, 1].hist(y_prob[y_test == 1], bins=50, density=True, alpha=0.5, color='#DC3545', edgecolor='white', label='Default') axes[1, 1].axvline(x=0.5, color='gray', linestyle='--', linewidth=1, label='Threshold = 0.5') axes[1, 1].set_title('Predicted Probability Distribution', fontweight='bold') axes[1, 1].set_xlabel('Predicted P(Default)') axes[1, 1].set_ylabel('Density') axes[1, 1].legend(loc='upper center', bbox_to_anchor=(0.5, -0.15), ncol=3, frameon=False) plt.tight_layout() save_fig('ch10_scoring') print("\n" + "=" * 70) print("CREDIT SCORING ANALYSIS COMPLETE") print("=" * 70) print("\nOutput files:") print(" - ch10_scoring.pdf: 4-panel credit scoring analysis")