Prediction of E-shopper´s Behavior Changes Based on Purchase Sequences

With the rapid development of online shopping, on-line one-to-one marketing becomes a great assistance to e-shoppers. One of the most important marketing resources is the prior daily transaction records in the database. In this study, the paper propose a new methodology for predict e-shoppers´ purchase behavior that uses e-shoppers´ purchase sequences. First, transaction clustering is conducted, then it is made that detecting the evolving e-shopper purchase sequences as time passes, and the e-shoppers behaviors, which are derived from a change in the cluster number of each e-shopper, are kept in the purchase sequence database. Finally, sequential purchase patterns over user-specified minimum support and confidence are extracted by using the association rule. The sequential purchase patterns are then stored in the association rule database. The better result is achieved by applying the new methodogy to a given example for e-shoppers.