A Combination Model to Predict Customer Needs in Online Book Sales

In this study, a method was presented based on data mining and Kano model to predict the type of online book stores customers' needs. First three groups of needs and factors affecting customer satisfaction of AdineBook online store were extracted, according to expert opinions and Kano questionnaire was designed based on these factors. After data preprocessing, type of customer needs were determined based on the Kano model. So customers were clustered according to type of needs and demographic characteristics. Clustering was done using K-means algorithm and the number of initial clusters were determined by self-organizing neural network. Then the frequency of customer needs in each cluster were prioritize and the clusters were classified accordingly. In the next step, decision tree technique was used to predict future customer needs. According to the result of decision tree, among the demographic variables as education, age, gender and marital status have the greatest impact on determining the type of customer needs. Originality/value: Considering the accuracy of decision tree that is too good, the results show that the proposed method has good efficiency to predict the needs of new customers