Clemson University
Experimental Design and Analysis for Decision Making: Methodology and Applications
Abstract
dc:description.abstractThis dissertation develops and applies advanced statistical and optimization frameworks to enhance decision-making under uncertainty, particularly in engineering and manufacturing contexts. First, we introduce an approach for the optimal design of controlled experiments that accounts for observational covariates, enabling more precise and personalized decisions. Second, we explore the application of constrained Bayesian optimization, using Gaussian process surrogate models, to optimize composite cure processes, significantly reducing computational effort while maintaining high predictive accuracy. Building on this foundation, we extend Bayesian optimization to bivariate Gaussian process models that capture correlations between objective and constraint functions, offering new insights into multidimensional decision landscapes. Finally, we propose a decision uncertainty quantification framework that integrates polynomial surrogates for simple cases and Gaussian process modeling for complex cases to estimate the variability in optimal decisions robustly. We validate these methodologies through case studies on cure process optimization and injection molding, demonstrating notable improvements in accuracy and efficiency over traditional approaches. The results indicate that these methods are well suited for larger-scale and more complex applications where systematic exploration of high-dimensional design spaces is required. Beyond theoretical insights, this work provides a practical roadmap for data-driven optimization in real-world settings. It highlights the value of personalized experimentation, robust surrogate modeling, and uncertainty-aware decision strategies. Future directions include expanding these frameworks to multi-objective problems, exploring non-separable covariance functions in Gaussian process models, and scaling to even higher-dimensional domains. Ultimately, the techniques presented in this dissertation aim to bridge the gap between rigorous statistical theory and complex industrial challenges, offering versatile tools for researchers and practitioners alike.
Degree
thesis:*- Name thesis:degree_name
- Doctor of Philosophy (PhD)
- Level thesis:degree_level
- Dissertation
- Discipline thesis:degree_discipline
- Mathematical Sciences
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Li, Yezhuo
- Contributors dc:contributor
-
- Qiong Zhang
- Whitney Huang
- Deborah Kunkel
- Christopher McMahan
Subjects
dc:subject × 9Identifiers
dc:identifier.*- Repository record dc:identifier
- https://open.clemson.edu/all_dissertations/4046