Identifying behavioural changes in movement path data

• Main Author: Knell, Andrew Stephen
• Published: University of Essex 2011
• Subjects: 519
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.573744

Movement path data can contain information that is vital for solving and understanding numerous biological and ecological problems. This thesis researches the problem of identifying behavioural changes in movement path data, in particular when there is two underlying contrasting behaviours. One behaviour is associated with relatively fast movement with small turning events and the other is associated with relatively slow movement with large turning events. Existing mathematical models that have been built to analysis such movement path data will be reviews in this thesis. Research has revealed that these models have limitations that restricts the accuracy of the movement path analysis. This was my motivation for creating two new models for identifying behavioural changes in movement path data: the Partial Sum Model and the Performance-Estimated Hidden Markov Model. The performance of these models will be tested and compared to existing models using simulation experiments. The results of these experiments provide strong evidence that the Partial Sum Model and Performance-Estimated Hidden Markov Model can perform better than existing models and it is conclusive that these models should be considered for analysing movement path data.