A new two-stage approach to solve the Linear Shelf Space Allocation Problem

In previous work we gave the first steps towards the development of a commercially viable tool that used evolutionary computation to address the Linear Shelf Space Allocation Problem (LiSSAP). This consists on allocating lengths of shelves in a given shop to specific products or groups of products. Here we present a procedure to solve the problem based on two stages. The first one relies on finding the "ideal" number of modules for every group of products. The second one uses an evolutionary algorithm (mu + lambda)-ES to fulfill groups affinity requirements. Moreover, a new distance measure between modules is also introduced, which has an important effect on the fitness function. Finally, we analyse different choices that can be made in the evolutionary algorithm and how they affect the global performance of the algorithm. These involve the chromosome encoding and the lambda parameter. The different combinations described are tested on two real problem configurations.