ML & Forecasting · Retail
AI Pricing & Promotion Decision Engine
Commercial teams must decide what to promote and how deep to discount, with uncertain demand response.
- My role
- Solo project.
- When
- 2025
+6.8%
simulated gross margin
11.2%
demand forecast MAPE
240 × 12
SKUs × stores simulated
18 mo
of synthetic sales history
Approach
Synthetic retail environment, time-aware model benchmarking, demand-response prediction, counterfactual simulation, and constraint-aware ranking of promo actions under a margin floor.
Results
On 18 months of synthetic sales for 240 SKUs across 12 stores, the engine's recommendations lifted simulated gross margin by 6.8% versus a rule-based promo calendar. Demand forecasts reach 11.2% MAPE with time-aware cross-validation.
Architecture
Technical highlights
- 01Synthetic retail environment with seasonality, promo effects, and price sensitivity
- 02Regression families benchmarked with time-aware cross-validation
- 03Constraint-aware decision engine with counterfactual scenarios and ranking
Stack
Related case studies
Quant Risk · MLOps
EVT Pre-Trade Limits Recalibration
94%
precision at a 1.3% false-positive rate
AI Engineering · Deep Learning
GPT from Scratch
10.8M
parameter decoder-only transformer, built from zero
ML & Forecasting · Sustainability
Agricultural CO₂ Emissions Modeling
68%
of emissions variance explained by two drivers