793: Bayesian Methods and Applications — with Alexandre Andorra

PyMC Framework

The PyMC framework is a powerful tool for Bayesian modeling, offering a range of features for both beginners and advanced users. Alexandre Andorra highlights the importance of choosing the right probabilistic programming language (PPL) based on one's proficiency in programming languages like R or Python 1. He suggests starting with tools like BRMS for R users or Bambi for Python users, which simplify the modeling process by making many decisions automatically 2. For those ready to dive deeper, PyMC and its Python counterpart, PyStan, offer robust options for more complex models. Andorra emphasizes the need to find a framework that resonates with the user, as "the models always fail unless it's the last one" 2.

   

ArviZ Visualization

ArviZ is an essential tool for visualizing and diagnosing Bayesian models, providing comprehensive distributions for all parameters. Alexandre Andorra explains that ArviZ is platform agnostic, meaning it can be used with models run in PyMC, PyStan, or other PPLs, as long as they output the required inference data object 3. This flexibility allows users to diagnose model convergence issues and present results effectively to clients. Andorra notes that ArviZ is available in both Python and Julia, making it a versatile choice for post-modeling workflows 4.