Assortment Planning

Assortment planning is one of the two classical OR problems in retail (the other is shelf-space allocation). From a candidate SKU pool, given a shelf or category page limit, the subset that maximizes expected margin or revenue is selected. Modern formulations model customer choice behavior with discrete choice models like the multinomial logit (MNL) or nested logit, so the substitution effect (probability of switching from X to Y if X is missing) is captured. Single-store / single-category problems use MIP or greedy assortment algorithms; multi-store uses Lagrangean decomposition or store clustering. The classical reference is Kök, Fisher, Vaidyanathan (2008); the MNL foundation is Talluri and van Ryzin (2004).