Robustness and applications to biology

点击翻译按钮获取中文摘要

In this proposal, I plan to continue collaborating with Biologists andMolecular Biologists on problems with substantive statistical issues. In all cases it is important to have procedures which arerobust against model misspecification. In the area of animal tracking data, the approach is to model the data using nonlinear state space models in which the errors are typically non-normal. In related research, issues involved in assessing trends in ecological series will be addressed. Both bootstrap and saddlepoint methodology will be developed to provide reliable inference for the proposed models. A related issue is to have methodology for model selection with particular interest in the case where the true model does not lie within the model frame.In molecular evolution, the focus will be on using the spectral envelope for sequences of amino acids with a view to classifying proteins as well as constructing phylogenetic trees.