MATLAB-AMPL integration with heuristics and association mining: An optimization-driven framework for retail shelf space allocation

This paper introduces a MATLAB-based framework for shelf-space optimization in retail, integrating AMPL via an API for solving linear and nonlinear programs. It supports two strategies: (1) Guided Random Rearrangement, creating constraint-aware layouts without prior data, and (2) Data-Driven Rearrangement, using association rule mining and mixed-integer programming to boost impulse buys. Core data structures are developed in MATLAB, while CPLEX and BARON solvers are employed through AMPL when required. The system adapts to varied retail environments, enhancing profitability and customer experience. New experiments with the Microsoft FoodMart dataset show that the data-driven method consistently outperforms random strategies. • This paper describes software for retail shelf-space optimization. • The framework integrates MATLAB and AMPL. • Supports constraint-aware random assignment and impulse purchasing optimization. • Customizable optimization criteria adapt to varied store formats and retail goals. • Empirical tests on Microsoft FoodMart data validate multi-period layout improvements.