A reduced variable neighborhood search-based hyperheuristic for the shelf space allocation problem

Abstract

The shelf space allocation problem (SSAP) aims to determine the optimal mix of product displays to maximize profitability. Decisions on the right product at the right location with the right space allocation are necessary when shelf space is limited. We adapt a comprehensive SSA model that considers own-space and cross-space elasticities, and conduct an extensive numerical study. We first implement the minimum shelf space requirement for each product and then develop HyRVNS, a pure random, Reduced Variable Neighborhood Search-based hyperheuristic framework to solve the problem. The strong potential of HyRVNS as a high-level heuristic is evidenced by the average percentage gaps of 0% to 0.92% at an average runtime of less than 1.0 s. Most importantly, for large instances, the numerical study shows that the proposed HyRVNS performs better at handling the problem instances altogether in both fitness and stability than do independently-implemented bespoke low-level heuristics.