Workshop on Statistical Methods for Dynamic
Vancouver, June 4-6 2009
System Models
Population dynamic time series in ecology tend to by short, noisy and to represent a huge amount of effort to collect. At the same time, the dynamic processes underlying the time series can be highly non-linear, and stochastically driven, making statistical inference about such processes difficult. This talk will discuss a pragmatic approach to inference with non-linear process based models of ecological time series, in which approximate likelihoods are developed based on simulating useful statistics of the series, and these likelihoods are then explored by MCMC.
Pragmatic inference for short ecological time-series
