Family
XGBoost + LightGBM + LogisticRegression stacked ensemble
in-developmentMITtabularb2b-facing
Credit Risk Scorer.
Probability-of-default credit risk model — stacked ensemble (XGBoost + LightGBM + LogisticRegression)
Foundation · XGBoost + LightGBM + LogisticRegression stacked ensemble
Marketing claim
Credit-risk scores with SHAP-level explainability and Courtroom signatures.Note ·The borrower receives a plain-language explanation of any decision (supporting the DPDP §11 access right); SHAP attributions are operator-facing and are translated into plain-language reasons in the borrower-facing portal.
What it does
The honest description.
Probability-of-default credit risk model — stacked ensemble (XGBoost + LightGBM + LogisticRegression)
Architecture
How it works.
Borrower features: debt-to-income ratio, payment history, bureau signals, income stability. Time-based train/test split to prevent lookahead bias. Probability of financial stress in next 90 days (0-1). NOTE: a deterministic debt-to-income heuristic (model_version=heuristic-dti-v1) currently serves; the SHAP-explained, fairness-audited (disparate impact > 0.80 across gender/region) stacked ensemble is the intended target, not yet served.
Training data & drift
What it learned from, and how we watch it.
Source
Inputs: Borrower features: debt-to-income ratio, payment history, bureau signals, income stability. Time-based train/test split to prevent lookahead bias.
Drift baseline · current quarter
Drift baseline not yet documented for this model.
What it's NOT
- Stacked ensemble (XGBoost + LightGBM + LogisticRegression) with 5-fold CV on base models.
- Fairness constraints + decision lineage tracked.
- Advisory — not a binding lending decision; final adjudication is the partner-lender's responsibility; Mukodi output is advisory.
Who calls this model
Across the house and the API.
Internal routes
No public routes documented yet.
Featured in
/features/credit-risk/features/b2b-api-sdksVersion history · immutable once published
Every release, traceable.
Data Mukodi holds about you is subject to DPDP §12 erasure (use /consent in the app), and any concern about this model can be raised via [email protected] under DPDP §13 grievance, or via /security.