Labs

Three “Lite” sandboxes that demonstrate decision-grade systems (not toy notebooks): Forecast → Price → Decide.

Forecasting Sandbox (Lite)

Purpose: Benchmark forecasting methods for SKU portfolios and select a defensible per-SKU model under demand-structure differences.
Output: Best model per SKU + evidence tables (MAE, Bias, Score = MAE + |Bias|) + regime tags (ADI/CV²) + audit ledger.
Link: Open Forecasting Sandbox

Pricing Decision Lite

Purpose: Choose a deployable price under elasticity uncertainty (not just a point-estimate “optimal” price).
Output: OPTIMIZE / HOLD / NO-GO decision + robust price recommendation + profit distribution evidence.
Link: Open Pricing Decision Lite

Decision Kernel Lite

Purpose: Make decisions under uncertainty using explicit assumptions (actions × scenarios × probabilities × losses).
Output: Recommended action + evidence tables for Expected Loss, Minimax Regret, and CVaR (tail risk).
Link: Open Decision Kernel Lite