dice-ml

Explainable, counterfactual & GenAI pipeline that turns sleep & lifestyle data into actionable stress-reduction advice — Predict (CatBoost/RF/TabNet) → Explain (SHAP) → Prescribe (DiCE) → Naturalize (GPT-4o-mini).

Lead scoring model-training lab with Litestar, Pydantic v2, SHAP drivers, and DiCE counterfactual planning.

TradingView strategy risk lab with model training, SHAP/DiCE explainability, Pydantic v2 validation, and Litestar API.

Prescriptive churn analytics with calibrated risk, uplift evidence, SHAP, counterfactuals, expected-value decisions, and a Next.js operations dashboard.

Explainable, counterfactual-driven decision support system for credit risk prediction. Combines SHAP/LIME-based risk scoring with a counterfactual recommendation engine for rejected applicants, subgroup fairness auditing (demographic parity, equalized odds), and an interactive dashboard for applicants and loan officers.

This is a XAI-powered heart disease risk predictor using SHAP, LIME, PDP & DiCE counterfactuals-deployed on Streamlit

Explainable ML pipeline that clusters feasible counterfactuals and distills them into global interpretable rules using DiCE, SHAP, Polars, DuckDB, and scikit-learn.