Business Value through XAI
summary
Retail demand-forecasting pipeline produces 44k weekly forecasts for four‑ and 20‑week horizons across 70,000 SKUs (262 categories) using >200 features, short‑term LightGBM and long‑term Prophet; model accuracy averages 74% (1−WAPE). An XAI layer uses SHAP values, counterfactual analysis, LASSO and rule/tree white‑box surrogates, surfaced in a weekly dashboard. Outcome: restored trust, higher automated order rates and product availability, and a measured 10% warehouse cost drop.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:14Speaker intro
- 2:39Demand forecasting project
- 8:05Model adoption challenges
- 10:15Explainability challenges
- 12:07XAI for business adoption
- 14:17XAI tools and techniques
- 16:04XAI dashboard
- 17:53Business impact
- 19:41Next steps
related talks
30mAI in Retail - Advancing Promo Demand Forecasting
35mBuilding a successful inventory planning solution with Amazon Forecast
09mDemand Forecasting in Supply Chain Using AI Basis Machine Learning
38mOptimizing Retail Forecasting Through AI Models
27mOptimizing Omnichannel Order Fulfillment with AI and Advanced Analytics
27mAI-Powered Pricing: Smarter Decisions, Bigger Profits
27mLeveraging AI for Resilient Supply Chains
50mTake AI/ML from theory to practice with Intel technologies