Demand Transfer Estimation at Scale via Restricted Logit Modeling

A new arXiv paper (2608.12680) proposes a method for estimating Demand Transfer (DT) coefficients at scale using restricted logit modeling. The research addresses inefficiencies in item demand forecasting for large item universes in retail assortment optimization. Traditional approaches require separate demand forecasts for every possible assortment, which becomes impractical with many categories. The alternative approach combines independent item demand forecasting with adjustments that account for relationships between items, central to which are DT coefficients representing the percentage of a target item's demand that transfers when substitute items are unavailable. The paper likely introduces a scalable estimation technique, though specific results are not detailed in the abstract. The work is relevant to retail analytics and operations research, potentially improving assortment planning efficiency. The paper is categorized as a cross-type announcement, indicating it may have been presented at a conference or journal. No specific authors, institutions, or locations are mentioned in the provided content.