21m1n / 2400-ZEWW752-Credit-Risk-Modeling

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Credit Risk Modeling Using WOE

Dataset: https://www.kaggle.com/teejmahal20/airline-passenger-satisfaction

  • Part I: Credit Scoring using WOE
    • Fine classing & calculate WOE
    • Coarse classing & calculate WOE
    • Logistic Regression
    • Quality Assessment

distribution of the scores card [1: defaulted / 0: not defaulted]

  • Part II: Tree-based Models

    • decision tree model
    • random forest model
    • model interpretability using SHAP
  • Part III: Model Comparison

models train_aucs test_aucs
1 logistic regression 0.95618 0.95890
2 basic decision tree 0.90282 0.90626
3 optimal decision tree 0.96584 0.96168
4 random forest 0.99167 0.99190

About


Languages

Language:R 100.0%