Hierarchical Bayesian Analysis of Peruvian Tree Growth Rates

This thesis explores the use of Bayesian statistical methods and hierarchical modeling in order to analyze massive data sets of Peruvian tree growth data. The study is important in finding connections between different parameters (such as the tree's classification or elevation) and rate at which the tree grows. We combine the Bayesian paradigm with the hierarchical structure of our data to make important inferences concerning contributions to tree growth. Using statistical simulation software, we sample continually from joint posterior distributions, updating our base prior assumptions in order to find posterior information for us to use in comparing different subgroups of the trees, as well as the effects of elevation. This allows us to set up various possible situations, then run our simulations and observe the differences which result. Nine different models were run and analyzed, taking different levels of tree classification into account, as well as different forms of elevation effect. The results allowed us to select the model which best informed us as to the potential growth rate of various trees.

Grantor dc:publisher

Wake Forest University

Year dc:date.issued

2011

Author dc:creator

• DeBenedetto, Joshua Albert

Language dc:language.iso

en

Handle dc:identifier.uri

http://hdl.handle.net/10339/33434

OAI identifier oai:identifier

oai:wakespace.lib.wfu.edu:10339/33434