Probabilistic programming: the Monte Carlo boost

Abstract

[en] The main goal of this project is to demonstrate the existing symbiosis between Monte Carlo Markov Chain (MCMC) methods and Probabilistic Programming. By carrying out a thorough study of how MCMC methods work, including an analysis of their algorithms and convergence, the in and outs of Probabilistic Programming can be better understood. One of the many applications of Probabilistic Programming is then employed in the study of the performance of a professional basketball player during his career. The model has been implemented using PyMC3, a Python package for sampling data using Monte Carlo Markov Chain methods.