Data Science ML E-commerce

Customer Segmentation & Churn Prediction

Built customer analytics system with RFM segmentation, churn prediction (AUC 0.91), and collaborative filtering recommendation for Indonesian e-commerce platform.

Python Pandas Scikit-learn LightGBM Surprise Power BI
2M+
Transactions
AUC 0.91
Churn Model
5
Customer Segments
78%
Precision

🎯 Key Features

👥 RFM Segmentation

Built Recency-Frequency-Monetary analysis identifying 5 customer clusters: Champions, Loyal, At Risk, Potential, Lost.

🔮 Churn Prediction

Trained LightGBM model achieving AUC 0.91 with 78% precision. Features include engagement score, order velocity, support tickets.

🎯 Recommendation Engine

Collaborative filtering using implicit feedback with matrix factorization (SVD). 15% CTR improvement in A/B test.

📊 Business Intelligence

Power BI dashboard with churn risk scores, segment trends, cohort analysis for marketing team insights.

📊 Customer Segments

⚙️ Churn Prediction Model

# LightGBM Churn Prediction import lightgbm as lgb from sklearn.model_selection import train_test_split features = ['recency_days', 'frequency', 'monetary', 'avg_order_value', 'support_tickets', 'email_opens'] X = df[features] y = df['churned'] X_train, X_test, y_train, y_test = train_test_split(X, y) model = lgb.LGBMClassifier( n_estimators=500, learning_rate=0.05, max_depth=6, num_leaves=31 ) model.fit(X_train, y_train, eval_set=[(X_test, y_test)], callbacks=[lgb.early_stopping(50)]) # Results: AUC 0.91, Precision 0.78