Clustering Customer Transactions: A Rough Set Based Approach

An efficient customer behavior analysis is important for good Recommender System. Customer transaction clustering is usually the first step towards the analysis of customer behavior. Traditionally data mining techniques are deployed in order to provide effective recommendation based on large population of customer transactions in real time. Customer transactions are likely to be imprecise and incomplete. Further the transactional dataset are normally dominated by heterogeneous and hierarchical product category preferences. Also the sparsity problem in customer transactions analysis is a big challenge. In this paper, a rough set based clustering approach has been proposed to cluster imprecise customer transactions in the presence of heterogeneous and hierarchical taxonomy of the products. It can also reduce the sparsity problem while analyzing the customer transactions. The algorithm can be used efficiently where a data warehouse is the knowledge source.