Development of an integrated intelligent product assortment optimization model for apparel retailing

Abstract

Product assortment planning of today's apparel retail industry mainly rests on the experience and subjective assessment of the decision maker in the company. Facing the increasingly fierce competition and fast changing customer demand, apparel retailers have stringent demands for optimizing product assortment by using intelligent and effective methods. The purpose of this research is to develop data mining-based methodologies for product assortment planning of apparel retailing. An effective framework for product assortment planning in an apparel retailing company was developed through integrating three types of problems, namely customer segmentation, customer preference identification, and product assortment optimization. On the basis of data mining technologies, these three types of problems were formulated mathematically and solved by effective methodologies. Clustering analysis was used to segment customers based on their behaviouristic data. A novel fuzzy c-means algorithm was proposed to tackle the issues of obtaining the optimal value of the fuzzy weighting exponent m and selecting the appropriate number of clusters in customer segmentation. After applying the method to the RFM data about customers, apparel retailers can obtain more reasonable and valuable results of customer segmentation as the method takes more factors (recency, frequency, and monetary) into consideration and uses the intrinsic characteristics of the data. The rough set approach was used to identify customers' preferred attributes. A weighted-incorporated rule identification algorithm was developed to solve the augmented formulation of rough set rule reduction. By employing this method, decision rules can be extracted for each homogeneous cluster of data records and relationships between different clusters. Considering different customer segments and customer preferences, an improved practical model based on an underlying multinomial logit (MNL) choice model for customers' selection of products is developed for optimizing retailers' expected profits from customers with heterogeneous preferences. The model provides retailers with a basis for several strategic decisions, including: (1) the optimal set of products offered in the market and their estimated sales; and (2) customers' preference structure influencing the optimal assortment with corresponding expected profits. Based on the historical transaction data from the local apparel retail company, experiments were conducted to evaluate the performance of the proposed methodologies. The experimental results demonstrate the effectiveness of the proposed methodologies for the apparel product assortment planning.