Evolutionary support vector regression for monitoring Poisson profiles

Abstract Many researchers have shown interest in profile monitoring; however, most of the applications this field research are developed under assumption normal response variable. Little attention has been given to monitoring with non-normal variables, known as general linear models which consists two main categories (i.e., logistic and Poisson profiles). This paper aims monitor problem Phase II develops a new robust control chart using support vector regression by incorporating some novel input features evolutionary training algorithm. The method is quicker detecting out-of-control signals compared conventional statistical methods. Moreover, performance proposed scheme further investigated for profiles both fixed random explanatory variables well non-parametric profiles. revealed be superior its counterparts, including likelihood ratio test (LRT), multivariate exponentially weighted moving average (MEWMA), LRT-EWMA other machine learning-based schemes. simulation results show superiority nearly all situations while it not able best simulations when there variables. A diagnostic learning approach also used identify parameters change profile. It that diagnosis reach acceptable comparison competitors. real-life example provided illustrate implementation charting scheme.