Type of Document	Dissertation
Author	Perry, Marcus Bradley
URN	etd-06032004-164842
Title	Robust Change Detection and Change Point Estimation for Poisson Count Processes
Degree	Doctor of Philosophy
Department	Industrial and Manufacturing Engineering, Department of
Advisory Committee	Advisor Name Title Joseph Pignatiello Jr. Committee Chair Anuj Srivastava Committee Member Chuck Zhang Committee Member James Simpson Committee Member
Keywords	Maximum Likelihood Estimation Hypothesis Testing Quality Control Special Cause Identification Statistical Process Control Poisson Count Processes Process Improvement Change Point Estimation Likelihood Ratio Change Point Detection Average Run Length Order-Restricted Inference CUSUM Control Chart PAV Algorithm
Date of Defense	2004-05-28
Availability	unrestricted
Abstract Poisson count process are often used to model the number of occurrences over some interval unit. In an industrial quality control setting, these processes are often used to model the number of nonconformities per unit of product. Current methods used for monitoring and estimating changes in Poisson count processes assume that the magnitude and type of change are known a priori. Since rarely in practice are these known, this dissertation reports on the development and evaluation of several methods for detecting and estimating change points when the magnitude and type of change are unknown. Instead, the only assumption requires that the type of change belongs to a family of monotonic change types. Results indicate that the methodologies proposed throughout this dissertation research provide robust detection and estimation capabilities (relative to current methods) with regard to the magnitude and type of monotonic change that may be present.
Files	Filename       Size       Approximate Download Time (Hours:Minutes:Seconds)     28.8 Modem   56K Modem   ISDN (64 Kb)   ISDN (128 Kb)   Higher-speed Access    mbp_manuscript.pdf 3.51 Mb 00:16:16 00:08:22 00:07:19 00:03:39 00:00:18