information-value

Find missing values in data set using Euclid distance, normalization and calculating information value, weight of evidence

Automatic optimal discretization pipeline

Weight of Evidence Encoding & Information Value

💰 Credit Risk Scorecard — Gradient Boosting + Logistic Regression + Decision Tree on 5,000 loans. Industry-standard metrics: Gini 0.521 · KS 0.395 · AUC 0.761 · IV/WoE table · Credit grades A-E. Basel III aligned. Production-realistic metrics. Python · scikit-learn

Bank-grade credit risk scorecard engine featuring Weight-of-Evidence (WOE), Information Value (IV), and adverse-action reason codes.

This repo contains algorithms for data analysis required while building DL/ML models

Interpretable WOE/IV credit scorecard on 150K borrower records — leakage-controlled validation, Optuna-tuned benchmarks, and policy analysis at matched approval rates

IFRS 9 Credit Risk Scorecard & Expected Credit Loss (ECL = PD * LGD * EAD) Engine under Basel III / EBA standards. Features R Weight of Evidence (WoE) binning & Information Value, Python PD models (Logistic Regression Gini=0.7467 vs XGBoost), 3-Stage Staging, PostgreSQL, automated Excel financial models, and a 2-page Power BI Dashboard.

Problem statment about modeling target vector and attempt to improve metrics