ENBIS: European Network for Business and Industrial Statistics

A maximum likelihood estimation of the multi-stage process change points

23 September 2009, 11:55 – 12:15

Abstract

Submitted by

Mehdi Davoodi

Authors

Mehdi Davoodi, Seyed Taghi Akhavan Niaki, Elnaz Asghari

Affiliation

Sharif University of Technology

Abstract

In most real-world manufacturing systems, production of goods comprises several auto-correlated stages while quality characteristics of goods at each stage are correlated random variables. In these systems, having an estimate of the process change point followed by a control chart signal would be useful to process practitioners. In this paper, a first-order autoregressive (AR(1)) is first employed to model multi-stage processes, and then the change point of this process is estimated through maximizing the likelihood function. We show that not only the proposed method well identifies the out-of-control stage, but also helps to locate the sample responsible for such departure.

Keywords: Change point estimation; multi-stage process, maximum likelihood estimation; autoregressive models

Return to programme