Nonparametric Change-Point Detection for Profiles with Binary Data

Abstract

Statistical process control (SPC) techniques for monitoring and diagnosing profiles have been proven to be important in many industrial applications. Profiles describe the relationship between the response variable and one or more explanatory variables. In some processes, the response variable of interest in profiles is binary, not numerical. The SPC problem for profiles with binary response data remains particularly challenging. Under such a premise, this article proposes a novel phase I scheme to detect the change-point in the reference profile dataset. The proposed method integrates change-point algorithm with the generalized likelihood ratio based on nonparametric regression, which could not only handle the generalized linear or nonlinear profiles, but also detect any types of changes in profiles. Numerical simulations are conducted to demonstrate the detection effectiveness and the diagnostic accuracy of the proposed scheme. Finally, a real example is used to illustrate the implementation of the proposed change-point detection scheme.