2014 Volume 27 Issue 2 Pages 81-93
We study the problem of web site classification for purchase or non-purchase customer estimation. A study on data mining of web site access log data is very important role in the field of e-commerce and web site design. In this study, we focus on the customer's behavior in the e-commerce web site. Firstly, we detect the difference between purchase and non-purchase customer's behavior based on the graph model. Secondly, we reconstruct the web site classification based on the graph model's result. Finally, we estimate the purchase or non-purchase customer's behavior using HMM by the reconstructed web site classification. As a result, our classification method shows better performance than other method.