Borrowing Strength Across Stores: Transfer Learning for Choice-Based Revenue Management

Key points

• Large retail chains must tailor assortment and pricing decisions to local customer preferences, yet even retailers with extensive sales records often have limited data for estimating demand at the store or local market level, particularly in product categories with short selling windows.1
• We study how retailers can balance these competing effects through transfer learning under a multinomial logit (MNL) demand model.1
• By exploiting its self-concordant-like properties, we establish finite-sample performance guarantees that characterize when information transfer improves demand estimation and how the appropriate degree of shrinkage depends on local sample size, feature dimensionality, and cross-store heterogeneity.1
• An empirical study using data from several hundred stores of a European apparel retailer shows that shrinkage improves out-of-sample product prioritization and revenue capture relative to both local estimation and full pooling.1

Sources (1)

• [1]Borrowing Strength Across Stores: Transfer Learning for Choice-Based Revenue Management
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 20, 04:08 PM

  We propose a shrinkage estimator that combines store-specific demand estimates with information pooled from other stores.

  Large retail chains must tailor assortment and pricing decisions to local customer preferences, yet even retailers with extensive sales records often have limited data for estimating demand at the store or local market level, particularly in product categories with short selling windows.

Extractive summary: sentences quoted from the sources.