27 Mar 2019
From data to decisions: an introduction to multiple criteria decision analysis, Ian Durbach, University of St.Andrews.
Seminar Room, The Observatory: 2:00 PM, 27 Mar 2019
RefID: 1973 click to edit (admin only)
This talk gives an introduction to the field of decision analysis. Complex decisions, like those faced in the management of environmental resources, involve finding a compromise between conflicting goals, often with uncertain or incomplete information about outcomes. Drawing boundaries around the decision can be difficult; sometimes what appears to be a single, once-off choice is actually a series of inter-related decisions. Different interest groups, each with their own opinions and preferences, are often involved.
Decision analysis provides a set of tools to deal with these kinds of challenges. Essentially this involves helping the decision maker to construct a model of their own preferences, which can then be used to guide the decision. In this talk I discuss this process of decision "aid": structuring the decision problem, developing mathematical models of preference from axioms of reasonable behaviour, and assessing the parameters of these models.
Seminar Room, The Observatory: 2:00 PM, 24 Apr 2019
RefID: 1988 click to edit (admin only)
Seminar Room, The Observatory: 2:00 PM, 22 May 2019
RefID: 1989 click to edit (admin only)
Point-transect distance sampling is particularly useful for surveying birds, and over the last four decades it is the standard method employed in Hawaiian forest bird monitoring. Accurate and precise estimates are essential to understanding how populations change across space and through time, particularly for rare species and in response to conservation efforts and management actions. Point-transect sampling involves recording distances to individually detected birds at samplers located along transects. In my previous CREEM seminar I demonstrated how the sampler-level data can be used to calculate density estimates and evaluate trends in densities for the endangered Hawai?i ?Akepa (Loxops coccineus) using smoother methods. In this presentation I detail how the detection probability uncertainty is propagated using posterior simulation methods and describe why methods such as bootstrap and delta methods are not appropriate. An alternative to density surface modelling and smoother methods is to model the density time series using state-space models. State-space models can be expanded to incorporate demographic vital rates. I describe the population dynamics models I have been exploring and illustrate how they improve inference by reducing uncertainty in the point-transect density estimates.
see also: Past Seminars