Customer Behavior Analysis using Web Usage Mining

Recent marketing strategies consider customers as important sources of the organization. Therefore, acquiring knowledge about customers and understanding their needs is necessary for keeping customers in an e-commerce business. Online customer shopping behavior is difficult to predict because they rarely visit the stores for real shopping, which is a challenge for marketers and researchers. Therefore, online business needs to analyze customers’ behavior in order to be successful. Hence, this study aims to provide a framework for increasing the accuracy of the analysis and recognition of customer groups as well as providing the model and rules for predicting customers’ behavior. Therefore, CRISP-DM and K-means algorithm were used for clustering data. Then, by assigning three tags of purchase, waiting and not purchase to customers, customers were categorized by C5 decision tree. Finally, a model with a precision of 63.6% and a collection of 261 rules with a confidence of 70% was obtained.