Optimal dynamic assortment planning with demand learning

摘要

Assessing customer preferences and offering products accordingly is crucial in retail business. For a retailer it may not be possible to display every product. The problem is then selection of products to offer. This is called the assortment planning problem. The situation is that retailers learns customer preferences over time and can improve on the current level of products offered continuously, thus making it a dynamic process. Fast fashion and online advertising are two examples of such a situation. There are three common features exhibited by assortment planning problems: 1. Customer purchase behavior may not be explicitly understood, 2. There may be substitutes for products but their profit generating capacity and demand vary, and 3. Assortment decisions can be dynamic. This study incorporates the above features in the stylized assortment planning problem and balances between information collection or exploration and revenue maximization or exploitation. The model does not take in to account, the pricing, inventory decisions, switching costs etc. The major constraint considered is the limited display capacity and that the prices are fixed throughout the selling season. (25 refs.)