Credit Risk — PD Modeling (Prototype)

Purpose: Prototype a Probability of Default (PD) model for retail personal loans.
Dataset: Task 3 and 4_Loan_Data.csv (columns include credit_lines_outstanding, loan_amt_outstanding, total_debt_outstanding, income, years_employed, fico_score, default).
Reference: CREDIT RISK ANALYSIS TASK.docx. :contentReference[oaicite:2]{index=2}

---

Contents

• notebook.ipynb — Jupyter notebook with ETL, EDA, Feature Engineering, Modeling (Logistic, RandomForest, XGBoost), and evaluation.
• Task 3 and 4_Loan_Data.csv — dataset (place in notebook folder).
• xgb_pd_model.joblib — trained XGBoost model (if saved).
• median_vals.joblib — medians for imputation (if saved).
• README.md — this file.

---

How to run

1. Install required packages (recommended to use a virtual env):

   pip install -r requirements.txt
   # or
   pip install pandas numpy scikit-learn xgboost matplotlib seaborn joblib