Analytics & Science for Customer Experience in E-commerce
summary
Flipkart builds SKU-level price intelligence for ~80 million products and ~150 million users, modeling demand with random forests/SVMs and unobserved component models, using MCMC for adaptive feature selection and a genetic algorithm for stochastic optimization of price vectors that account for cross‑elasticity/cannibalization and inventory/display effects. Models run near real time with guardrails for price trust, filter out low‑accuracy SKUs, and produce category‑level price recommendations; internal A/Bs show the optimizer improves the objective in at least one deployed category.
AI-written summary · about
chapters (7)
- 0:00Start
- 1:04Scale of Flipkart Business
- 8:53Methodology
- 11:51Demand-Price Response Function
- 13:20Factors Affecting Demand
- 15:30Demand Prediction Model Finalizing Process
- 17:32Optimization Strategy
related talks
06mINTERVIEW
55mImproving product discovery via relevance and ranking optimization
22mFrontiers of Machine Learning from the Lens of Retail
08mMachine Learning for Bangladeshi Retailers
27mAI-Powered Pricing: Smarter Decisions, Bigger Profits
27mOptimizing Omnichannel Order Fulfillment with AI and Advanced Analytics
01h 00mAutomation: Amazon’s Supply Chain Optimization Technologies (SCOT)
23mMachine Learning with a Deep Dive into Automated Forecasting System