Robust mixture modelling of telemetry data in wildlife studies of home range

Bruce Worton, C. R. McLellan

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper considers the advantages of using mixtures ofdistributions, rather than the standard nonparametric approaches, for estimation in wildlife studies of home range. Mixtures are only used in a very limited way currently even though they were first proposed for such modelling in 1983. However, recent advances in the theory and computational techniques for mixtures of distributions mean that they may be used to analyse radio telemetry data and the models used to extract important ecological features of animal behaviour. We illustrate that robust mixture modelling is necessary as a common feature of telemetry data sets is that they contain outliers which lead to problems when determining appropriate mixture models. In particular, we investigate the use of mixtures of bivariate t distributions to take account of outliers, and study their properties. An application to brush rabbit telemetry data is used to show the advantages of using such heavy tailed mixtures.
Original languageEnglish
Title of host publicationProceedings of the 26th International Workshop on Statistical Modelling
Editors Conesa, D., Forte, A., Lopez-Quilez, A. and Munoz, F.
Pages656-659
Publication statusPublished - 15 Jul 2011

Fingerprint

Dive into the research topics of 'Robust mixture modelling of telemetry data in wildlife studies of home range'. Together they form a unique fingerprint.

Cite this