This project showcases a full pipeline from data preprocessing, exploratory data analysis, model training (default, tuned, weighted, SMOTE), to SHAP-based interpretation.