Customer state estimation with Poisson distribution model

Abstract

In this paper, a new access log analysis is proposed which estimates both active states and inactive states from observations simultaneously. I improved burst analysis and developed enthusiasm analysis to detect not only active states but also inactive states. Speaking concretely, I constructed a generative model that assumed observations, which meant event occurrence frequency in this paper, were generated under some Poisson distributions. The generative model based on Poisson distributions consists of some distributions including $$\lambda$$?s which are less than average frequency and more than average frequency. A cost function, which consists of log likelihood and state transition cost, is defined and enthusiasm levels are estimated from the observations minimizing the cost function. The proposed method was applied to query occurrence data in access logs and I confirmed the proposed method could find active states and inactive states from submission patterns and the results showed relationship between enthusiasm levels and real events.