Introduction to Bayesian Model Evaluation, Visualization, and Comparison Using ArviZ
Ravin Kumar, ArviZ
Colin Carroll, Freebird
In this tutorial we will build your expertise in handling, diagnosing, and understanding Bayesian models. It is intended for Bayesian modelers that know how to fit models, but desire further understanding on model criticism and visualization techniques. We will cover how to work with model data, how to evaluate model fit and how to communicate results. Attendees learn how to 1. Evaluate and visualize models 2. Understand plots commonly encountered in Bayesian contexts Bayesian modeling expertise is not required. Knowledge of python syntax and Numpy/Pandas are helpful to complete activities in this tutorial. Even without coding experience attendees may find value in learning how to interpret Bayesian model diagnoses and visualizations created by others.
NumPy, SciPy, Pandas, Matplotlib. Basic familiarity with skills above will help students work with plots, and dataframes/ndarrays, but prior knowledge is not strictly required.