Thesis
Bayesian analysis of agricultural treatment effects in the presence of a fertility trend and outliers
Honours, Murdoch University
2005
Abstract
This thesis endeavors to study the Bayesian technique of making inferences, which was assisted by the use and study of the Markov Chain Monte Carlo (MCMC) approach and accompanied by the implementation of some models using the Bayesian program, winBUGS.
The models investigated in this thesis were based on a model used by Taplin and Raftery (1994). These models were concerned with estimating treatment effects in the presence of a fertility trend and outliers, where the response variable was crop yield and the other parameters were treatment effects, fertility effects and an error term.
The thesis includes a brief review of some underling principles concerning Bayesian analysis (Section 1) and MCMC (Section 2). It also includes a review of some literature that is related to the Bayesian estimation of treatment effects for agricultural data and techniques for accommodating outliers (Section 3). The practical section of the thesis (Section 4) was the estimation of treatment effects for some data. WinBUGS codes were written and its components verified and explored.
Further research into the robustness of two techniques for accommodating outliers was also explored. This being comparisons between the variance inflation technique and the response variable following a Student's t-distribution.
Details
- Title
- Bayesian analysis of agricultural treatment effects in the presence of a fertility trend and outliers
- Authors/Creators
- Christina Gracie
- Contributors
- Ross Taplin (Supervisor)
- Awarding Institution
- Murdoch University; Honours
- Identifiers
- 991005544008407891
- Murdoch Affiliation
- School of Engineering Science
- Language
- English
- Resource Type
- Thesis
Metrics
7 File views/ downloads
102 Record Views