Oft encountered models

We have thus far emphasized that model-building should be bespoke, which is to say that whatever models we build should be what we think best describes the data generation process, and not necessarily models that are well-known or with very convenient analytical forms of their maximum-likelihood estimates of p-values. Though, as stated in the Zen of Python, practicality beats purity, and sometimes those handle models and their established results are useful.

In the following lessons, we focus on three models, or classes of models, that come up very frequently in model building for statistical inference in the biological sciences. First, in lesson 41  Mixture models, we discuss mixture models. Then, in lessons 42  Variate-covariate models through 45  Implementation of MLE for variate-covariate models, we discuss variate-covariate models, which result in inference procedures commonly referred to as curve fitting. Finally, in lessons 46  Principal component analysis and 47  Implementation of PCA, we present principal components analysis.