Customer State Analysis with Enthusiasm Analysis

摘要

In this paper I propose customer behavior analysis using enthusiasm analysis, which estimates customers' activation levels. Finally our goal is to discovery drop-off users from access logs according to the activation level. It is important to find drop-off users earlier and pay attention to them to stay in a service. Usually prediction models are constructed with machine learning and RFM analysis is used in marketing field. In this paper I estimate enthusiasm levels, which denote customers' activation, from observations and apply them to prediction of discovery of drop-off users. In evaluational experiments I use real online shop access logs and discuss relation between enthusiasm levels and drop-off users. I confirmed that many drop-off users took lower enthusiasm levels in evaluation point and the enthusiasm level could be used to predict drop-off users.