Once a model is fitted to a dataset, it is hoped that new information is gleaned from the model that would otherwise not be possible with the data alone. Parameter identifiability is a common goal of such an analysis, although by itself may not be sufficient to reveal the informativeness of the dataset, or inform future data collection activities. This course will introduce the analysis of sloppiness as a tool for unveiling parameter uncertainty when mathematical models are fitted to data, and demonstrate examples in ecology where this tool can inform future data collection activities and/or inform future model designs.
1. Model-data calibration using Bayesian and frequentist approaches.
2. Visualisation of calibration outputs.
3. Analysis of model sloppiness.
4. Interpreting model sloppiness outputs, and connections to other data-informing approaches.
This course assumes an undergraduate level understanding of the principles of statistical modelling and inference (including likelihood functions). Some familiarity with systems of ordinary differential equations stability concepts, and numerical methods for solving ODEs. Tutorial will include some use of Matlab programming language, thus the course requires basic knowledge of Matlab and working software. However, the emphasis will be on exploring models, rather than coding, and templates will be provided.
This course will be relevant to anyone with an interest in mathematical modelling of natural processes, particularly in ecology,