期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2015
卷号：74
期号：3
出版社：Journal of Theoretical and Applied
摘要：Electronic supplier relationship management (e-SRM) is important in order to maintain strong, long lasting and beneficial relationship between e-commerce firms and their suppliers. One important function of e-SRM is to predict suppliers who tend to churn such that early �treatment� can be given. In the e-commerce systems that involve suppliers as the websites users, predicting suppliers� churn tendency can be based on analyzing their frequencies in accessing the e-commerce websites. Our proposed techniques include data warehouse design (supporting the data collection and preprocessing) and unsupervised algorithms that analyze the preprocessed bitmaps of time series data representing suppliers website access from time to time. Having bitmaps as inputs, our proposed algorithms are efficient (the time complexity is O(n)) as proven with our experiments. In experimenting with real world data of an e-commerce system selling hotel rooms, our techniques produce output of supplier segments where each segment has certain churn level tendency and need specific treatment.
关键词：Churn Prediction In E-Commerce; Supplier Relationship Management; Web Usage Mining
