Bank A aims to develop a forward-looking Behaviour Score—a classification model predicting whether a credit card customer will default in the following month. This project uses anonymized historical data of over 30,000 customers to build an interpretable and high‑performance credit risk model.

CreditCardProject/
├── .gitignore
├── README.md
├── CreditCardBehaviourScorePred.ipynb   # Jupyter notebook with full pipeline
├── submission_22112075.csv # Final predictions
├── Datasets_final/
│   ├── train_dataset_final1.csv         # Training data 
│   └── validate_dataset_final.csv       # Validation data 
├── requirements.txt                     # Python package dependencies
└── .venv/                               # Virtual environment

• Customer_ID: Unique identifier
• Demographics: sex, education, marriage, age, LIMIT_BAL
• Payment Status: pay_0 to pay_6 (last 6 months)
• Bill Amounts: Bill_amt1 to Bill_amt6
• Payment Amounts: pay_amt1 to pay_amt6
• Aggregates: AVG_Bill_amt, PAY_TO_BILL_ratio
• Target: next_month_default (1 = default, 0 = no default)

git clone https://github.com/payalkanyan/CreditCardBehPred
cd CreditCardBehPred
python -m venv .venv
source .venv/bin/activate        # Linux/Mac
.venv\Scripts\activate          # Windows
pip install -r requirements.txt

• Class balance and default rate (19%)
• Demographic trends (higher default among younger customers)
• Payment behaviour analysis (pay_0 correlation)
• Advanced EDA: correlation heatmap, trends over time

• Delay features: avg_delay, delay_count, max_delay, improvement
• Financial ratios: util_ratio, underpay_ratio
• Raw aggregates: bill_total, pay_total

• Train/test split (80/20) with stratify

• Handle class imbalance via SMOTE

• Models compared:

  • Logistic Regression (baseline & tuned threshold)
  • Decision Tree, Random Forest
  • XGBoost, LightGBM
  • StackingClassifier (LogReg+RF+XGB) with meta learner

• Evaluation metrics: F2-score (primary), AUC-ROC, Precision, Recall

• Threshold tuning for best F2 (optimal = 0.25)

1. Open CreditCardBehaviourScorePred.ipynb
2. Run all cells in order
3. Inspect results and threshold tuning
4. Generate final submission

Listed in requirements.txt:

pandas
numpy
scikit-learn
imbalanced-learn
xgboost
lightgbm
matplotlib
seaborn