Introduction to Bayesian modeling with PyMC3

Juan Orduz will speak about PyMC3 (https://pymc-devs.github.io/pymc3/) and Bayesian modeling. The aim of this talk is to give an introduction to PyMC3, a Python package for Bayesian statistical modeling and Probabilistic Machine Learning. In the Bayesian framework quantities of interest, such as parameters of a statistical model, are treated as random variables. One draws conclusions of these quantities by analyzing the posterior distribution obtained from observed data. In practice, the resulting posterior distribution does not always have a known parametrization and therefore sampling methods become very handy.

The outline of the talk is as follows: First we will briefly recall the basic principles of Bayesian modeling. Then, we will treat a concrete and simple example, using a Jupyter notebook, where we show how to set up a Markov chain Monte Carlo simulation with PyMC3.

This talk is addressed for non-experts and no prior knowledge on Bayesian modeling is required.

Lightning Talks

Usually there will also be the opportunity to give lightning talks (http://en.wikipedia.org/wiki/Lightning_talk). Please consider giving one and register yours on this EtherPad (https://pad.freitagsrunde.org/W45Lwmah9e)!

Restaurant

At about 8:45 pm we’ll move to a nearby restaurant.