Dissertation
MONITORING PRODUCT RELIABILITY WITH COMPETING RISKS USING REAL-TIME CONTRASTS AND STATISTICAL ANALYSIS OF BLUEBERRY PHENOLOGICAL STAGES
Washington State University
Doctor of Philosophy (PhD), Washington State University
01/2022
DOI:
https://doi.org/10.7273/000004352
Handle:
https://hdl.handle.net/2376/125001
Abstract
This dissertation includes two papers covering independent topics. The first paper discusses the application of the Real-Time Contrasts (RTC) method of process monitoring on competing-risk failure data. Lab test results are simulated to explore theperformance of this approach which is compared to that of logistic regression, another method of process monitoring. A method of feature selection as well as a way to handle fault detection is also discussed.
The second paper explores different approaches of modeling the phenological stages of blueberry plants. Different methods explored include ordinal logistic regression models, multi-state models, approximate latent linear models, random forests, and neural networks. Different feature configurations and stage structures are considered. Modeling methods are compared through the mean squared error of the stage predictions as well as the prediction probabilities.
Metrics
4 File views/ downloads
30 Record Views
Details
- Title
- MONITORING PRODUCT RELIABILITY WITH COMPETING RISKS USING REAL-TIME CONTRASTS AND STATISTICAL ANALYSIS OF BLUEBERRY PHENOLOGICAL STAGES
- Creators
- Robert Kelly Stancil
- Contributors
- Francis G Pascual (Advisor)Nairanjana Dasgupta (Committee Member)Yuan Wang (Committee Member)
- Awarding Institution
- Washington State University
- Academic Unit
- Mathematics and Statistics, Department of
- Theses and Dissertations
- Doctor of Philosophy (PhD), Washington State University
- Publisher
- Washington State University
- Number of pages
- 110
- Identifiers
- 99900883035201842
- Language
- English
- Resource Type
- Dissertation