Oksana Chkrebtii
My research is in the area of statistical inference for complex systems from experimental or observational data. I develop methodologies for parameter estimation and uncertainty quantification for dynamic systems models as well as stochastic generating mechanisms with mathematically intractable likelihoods. These play an important role in describing biochemical processes, animal movement, and the spread of invasive species, where the accuracy of prediction relies on inference on unknown system parameters. One aspect of my research is in characterizing numerical uncertainty in the solution of analytically intractable systems of partial and ordinary differential equations, distinguishing different sources of variability, and establishing their impact on estimation. We consider numerical solutions from a Bayesian perspective as an inference problem, which can then be incorporated directly into the inference process.
Another aspect of my work is in constructing estimation methods for processes described by stochastic mechanisms for which neither a closed-form likelihood nor a close approximation are available. Such methods have recently been introduced for use in genetics and epidemiology, but are relatively new in ecological applications. As part of a collaborative project, I am working on likelihood-free inference for modeling introduction and spread of non-native Earthworms in the Boreal Forest of Northern Alberta.