The Process State Tracking Method in Poisson Process

Understanding the state transition of a process from the time series data obtained from the process is important from the viewpoint of both analyzing and controlling the process. In particular, it is important to clarify a turning point of the state transition, that is, the point of change in the process and to find the cause of the state transition by observing and analyzing the data from the process. This paper considers the method of detecting several change points in a process based on the likelihood theory and information criterion when a series of data from the process follows the Poisson process. Then, the method of finding any process state fluctuation and points of change is called “the process state tracking method”. The validity and applicability of the process state tracking method introduced in this paper is confirmed through some numerical applications. © 2021 Japan Industrial Management Association. All rights reserved.