Growth Projections and Assortment Planning of Commodity Products Across Multiple Stores : A Data Mining and Optimization Approach

Product assortment and availability are important determinants of sales success for firms of industrial commodity products. Well-known pricing and promotion strategies for differentiated products do not translate well to such products where price is closely tied to the cost of the products. Consequently, firms with multiple stores of commodity products are faced with the problem of product assortment that incorporates varying geographic and demographic conditions of locations they serve. The paper presents a model for assortment planning and optimization for multiple stores of a company. The novelty of our approach is twofold: first, it deploys data mining techniques to identify sales pattern information across multiple stores through existing sales data across segments and across stores; second, it identifies the optimal product assortment for each store and permits analyses of assortment efficiency evaluation among all existing stores. Our model first finds frequent itemsets based on association rule analysis and prunes them using a novel conflict resolution method. It then incorporates the identified product combinations into the development of the optimization formulation. Our methodology offers solutions that have important implications on product assortment, including complements versus substitutes and product bundling, and sheds lights on product planning and assortment strategies in general. A data set from an industry leading plastics manufacturer and retailer in the United States is used to demonstrate our model