Model assessment

Thus far, we have not been terribly principled in how we have assessed models. By model assessment, I mean some metric, quantitative or qualitative, or how well matched the proposed generative model and the actual (unknown) generative model are. In my opinion, many of the most effective methods are graphical, meaning that we make plots of data generated from the theoretical generative model along with the measured data. We will nonetheless also explore information criteria for model selection.