Türkiye'deki bir süpermarket zinciri için ortak bir çeşit planlama ve raf payı dağıtma optimizasyonu

We consider a joint assortment planning and shelf space allocation problem. The assortment planning sub problem consists of introducing special products. If these special products are introduced, they can be placed on the shelves that were occupied previously by the standard products. The reduction in the shelf space allocated to the standard products can result in the decrease of the demands for these products. On the other hand, the introduction of the special products can increase the demand for the standard products through cross-selling. The problem is formulated through a nonlinear integer programming problem. The cross-selling effects are estimated by analyzing real sales data of one year of the biggest supermarket chain of Turkey. We derive a greedy heuristic procedure that can be used easily by practitioners. To compare the performances of our heuristic with one of the most used nonlinear integer programming solvers, namely Dicopt, we solved 100 medium-scale test problems by GAMS 23.7 (using Dicopt) and our heuristic. Our heuristic solved all of the test problems while GAMS solved 78 out of 100 test problems within 17 minutes time limit; i.e., if GAMS did not provide a solution within 17 minutes, we stopped the optimization procedure. For each problem, our heuristic outperformed GAMS in terms of the optimal objective value as well as the execution time. The optimal objective value found by our heuristic is, on average, 14% higher than the optimal objective value found by GAMS. The average execution time of our heuristic is less than one quarter of the average execution time of GAMS.