My Role
Data Scientist
Background
Credit Risk models needs to be up to date and validated every year
Data
Credit data from customers
Evaluation Metric
The evaluation metric used is Area Under the Receiver Operating Characteristic Curve (AUROC), a metric for credit scoring. Repeated Stratified K-fold split the data while preserving the class imbalance and perform k-fold validation multiple times.
Finding
ROC curve, PR curve, calculate AUROC and Gini. The AUROC on test set comes out to 0.866 with a Gini of 0.732, both considered as acceptable evaluation scores. The ROC and PR curves:
