AI in Retail - Advancing Promo Demand Forecasting
summary
Claims a 2.5-fold reduction in forecasting error for a 400+ store rollout by combining LightGBM with feature engineering that models cannibalization, gap-filling for sparse long-tail SKUs, and categorical grouping instead of raw SKU IDs. Outputs are distributional via quantile conformal regression so buyers pick tolerable risk levels. Hundreds of features are pruned to ~15% using wrapper methods like Boruta and SHAP. Production deployment runs on GCP APIs and starts with a 3–4 month pilot.
AI-written summary · about
chapters (11)
- 0:00AI in retail forecasting
- 1:12Pandemic demand challenges
- 3:00AI for leading retailer
- 4:01Data challenges in forecasting
- 6:55Promotions strategy
- 8:02Technical forecasting challenges
- 12:12ML techniques and innovations
- 14:46Promotional campaign decisions
- 17:57AI adoption factors
- 21:29AI implementation steps
- 22:46Q&A
related talks
38mOptimizing Retail Forecasting Through AI Models
20mBusiness Value through XAI
28mMastering Demand Forecasting: Lessons from Europe's Largest Retailer
07mAI in Retail
23mArtificial Intelligence for Store Assistance
16mGen AI For Retail Data Management
28mHow Human Expertise and AI drive Scalable Forecasting Success at
31mFrom stockouts to happy customers: Proven solutions for time series forecasting in retail