Variable Selection for Recommendation System of Purchase Timing for EC Site

Abstract

Recommendation is the one of the most representative marketing method for Electoronic Commerce.
Comparing with retail shop, EC can grasp visitors simultaneously as they visit, and not only the purchase record but also the access log of them.
We have proposed a method to grasp a sign of purchase using statistical method.
This method can estimate well, however these statistical method can not implement on existing recommendation system. So in this study, we propose an alternative method to discover the sign of purchase to implement on recommendation system.