Discovery of online shopping patterns across websites

摘要

Market basket analysis discovers customer purchasing patterns by extracting product associations from purchase transactions. This information helps store managers to make better decisions about shelf allocation, and product promotion. Though market basket analysis has met with success in individual stores, it has rarely been made in Internet purchase transactions, which often spans multiple sites. On a web store, the consumers can easily navigate to different online stores to make purchases or make the purchases from the same store. The analysis of online shopping data discovered a number of patterns of associated purchases from multiple sites. The research work focuses on this type of pattern i.e., the purchase of products on one site is associated with the purchase of products on other sites and the sites do not need to be specific. The problem has been introduced, followed by the definition of the online shopping pattern (OS-pattern). Two frequent OS-pattern mining methods, OSP-tree and OSP-level have been developed and compared. The two patterns have been experimented on real Web shopping data and synthetic data. The related work in various mining literature categories has been discussed.