Dissertation
DEPENDENT COMPETING RISKS MODELING USING MULTIVARIATE WEIBULL DISTRIBUTION AND PROCESS MONITORING WITH CLASSIFICATION TECHNIQUES AND REAL-TIME CONTRAST
Doctor of Philosophy (PhD), Washington State University
01/2019
Handle:
https://hdl.handle.net/2376/112178
Abstract
In this dissertation, two papers for two independent topics are included. In the first
paper, we construct a model for dependent competing risks. We discuss the model as
sumptions and the use of a multivariate Weibull distribution as the underlying model
for the latent times of the competing risks problem. We address the identifiability
issue and the use of proxy model for this problem. It has been shown that the in
troduction of parameters to model the correlation among latent times of risks is very
important. We propose three estimation methods for the model and apply them to a
real example and present the numerical results for each of the methods.
In the second paper, we study the process monitoring under the framework of
real-time contrast (RTC) which uses the most recent data to contrast with reference
data and to update the decision boundary. Building on existing work, we extend the
distance-based control chart with linear discriminant analysis (LDA) classifier to the
chart to quadratic discriminant analysis (QDA) classifier under the RTC framework.
The latter is proven to have better efficiency in detecting distribution shifts caused by
the changes in covariance or correlation structures. We also develop RTC chart using
logistic regression which is shown here to have the ability of detecting shifts in means
but not shifts in variances. The performance of the proposed charts are compared
with the existing RTC-RF and RTC-LDA chart with both simulation studies and a
real example.
Metrics
11 File views/ downloads
26 Record Views
Details
- Title
- DEPENDENT COMPETING RISKS MODELING USING MULTIVARIATE WEIBULL DISTRIBUTION AND PROCESS MONITORING WITH CLASSIFICATION TECHNIQUES AND REAL-TIME CONTRAST
- Creators
- Min Ye
- Contributors
- Francis Pascual (Advisor)Marc Evans (Committee Member)Yuan Wang (Committee Member)
- Awarding Institution
- Washington State University
- Academic Unit
- Department of Mathematics and Statistics
- Theses and Dissertations
- Doctor of Philosophy (PhD), Washington State University
- Number of pages
- 119
- Identifiers
- 99900581415601842
- Language
- English
- Resource Type
- Dissertation