An integrating approach to root cause analysis of a bivariate mean vector with a linear trend disturbance

Abstract

Identifying the change point or the starting time of a disturbance in a process can play an essential role in root cause analysis. Although, in univariate environment, change point identification by itself could be a significant step in process improvement but in multivariate environment without conducting diagnostic analysis which leads to the identification of the variable(s) responsible for the change in the process, one cannot effectively perform root cause analysis. In this paper, a supervised learning approach based on artificial neural networks is proposed which helps to identify the change point in a bivariate environment when the process mean vector is allowed to shift linearly and simultaneously a diagnostic analysis is conducted to identify the variable(s) responsible for the change in the process mean vector. To the best of our knowledge, this is the first time that this problem is addressed in the literature of change point estimation. The performance of the proposed model is investigated through several numerical examples. Results indicate that the proposed model provides practitioners with a very accurate estimate of the change point in the process mean vector and simultaneously helps to identify the variable(s) responsible with the out-of-control condition.