|
[Print-friendly version]
 Sandra Johnson
Faculty of Science, Statistical Science and Operations Research
Thesis Title: Bayesian Networks with applications in Ecology
Current Thesis Abstract: Bayesian networks (BNs) provide the means of modelling complex environmental problems within a sound statistical framework. The strength of BNs is that they are visual and able to combine existing data and expert opinions to model a wide variety of situations in a diverse field of applications. The visual nature of BNs effectively engages experts to capture their knowledge within the network. BN modelling is typically an iterative process and the model is refined as more information becomes available. The validity of the model can be verified against known cases. There are many software applications available commercially to build BNs. We are mainly using Netica® and Hugin®.
My proposed research will extend an existing static model for Lyngbya algal bloom initiation temporally to create a dynamic Bayesian network (DBN) and will explore static and dynamic Lyngbya BN models in more detail. The feasibility of creating a Lyngbya ‘meta model’ will be assessed, after doing a literature review of Lyngbya blooms worldwide.
Cheetahs are on the Red List of Threatened Species of The International Union for the Conservation of Nature and Natural Resources (IUCN). The current IUCN classification for the cheetah is Vulnerable, VU C2a(i), which means that it is considered to have a high risk of extinction in the wild. Using the techniques employed in the algal bloom model, a DBN will be built
to model the survival of relocated wild cheetahs
with the help of cheetah experts and an overseas statistical collaborator.
|
