Bayesian Analysis with Python

Summary

In this chapter, we built from the last two by extending our abilities to manage models with more than one parameter. This turned out to be something very simple to do with the help of PyMC3. For example, obtaining the marginal distribution from the posterior is just a matter of properly indexing the trace. We also explored a few examples of using the posterior to derive quantities of interest from it, such as synthetic data or measures to better explain the data. We found the Gaussian model for the first, but certainly not the last, time, since it is one of the pillars of data analysis. Before we had any time to glorify the Gaussian model, we pushed it to its limits with the help of potential outliers in the data. Therefore, we learned ...