Please use this identifier to cite or link to this item:
|Title:||EXTENSIONS TO THE EXPONENTIALLY WEIGHTED MOVING AVERAGE PROCEDURES||Authors:||CHAO AN-KUO||Keywords:||Change Detection, Statistical Process Control, Exponentially Weighted Moving Average, Maximum Likelihood, Hypothesis Testing, Markov Chain||Issue Date:||29-Dec-2016||Citation:||CHAO AN-KUO (2016-12-29). EXTENSIONS TO THE EXPONENTIALLY WEIGHTED MOVING AVERAGE PROCEDURES. ScholarBank@NUS Repository.||Abstract:||In change detection, persistent changes can be identified by change detection procedures. However, their practical use may be limited because the post-change distribution is usually unspecified. It may require a large sample size for estimating its parameters, resulting in a delay to detection. To address this problem, we develop multiple exponentially weighted moving average (EWMA) procedures in two scenarios. When the pre-change distribution is specified, we propose three new EWMA procedures based on the Lagrange multiplier test, the Wald test, and the gradient test. These procedures are asymptotically equivalent to the EWMA procedure based on the likelihood-ratio test. They can be applied at the arrival of the first observation if the distribution belongs to the exponential family. In addition, we extend the proposed EWMA procedures to the scenario where the pre-change is partially specified. These extensions are relatively robust against changes in the nuisance parameters.||URI:||http://scholarbank.nus.edu.sg/handle/10635/135829|
|Appears in Collections:||Ph.D Theses (Open)|
Show full item record
Files in This Item:
|Thesis.pdf||1.1 MB||Adobe PDF|
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.