Online Courses in Advanced Bayesian Modeling Techniques with Stan start in under two months

Publicity Events
1

Interested in courses on mixture, ordinal, survival or pairwise comparison modeling in Stan? Then read on. Courses started in under two months.

Despite the promise of big data, inferences are often limited not by the size of data but rather by its systematic structure. Only by carefully modeling this structure can we take fully advantage of the data – big data must be complemented with big models and the algorithms that can fit them. Probabilistic programming languages like Stan facilitating this modeling, allowing us to implement bespoke models and while providing state-of-the-art algorithms to compute Bayesian inferences.

In these courses, Advanced Bayesian Modeling In Stan - Powered by Eventzilla, Michael Betancourt presents a series of advanced Bayesian modeling techniques and their implementation in a principled Bayesian workflow, including discussions of prior modeling, inferential degeneracies, and more. Each course module incorporates interactive exercises run through RStan, the R interface to Stan, and PyStan, the Python interface to Stan.

The course consists of four modules each covering a different topic. Each module is are offered in parallel morning and afternoon (EST) sessions for scheduling flexibility and can be taken independently of each other. Modules are presented remotely through video conferencing and a dedicated Discord server, with all slides, recordings, and exercises made available to attendees.

Module 1: Mixture Modeling
Monday August 18, Thursday August 21

Module 2: Survival Modeling
Monday August 25, Thursday August 28

Module 3: Pairwise Comparison Modeling
Monday September 15, Thursday September 18

Module 4: Ordinal Modeling
September 22,Thursday September 25

For detailed module descriptions and course logistics see the course page at Advanced Bayesian Modeling In Stan - Powered by Eventzilla. Questions can also be addressed directly to courses [at] symplectomorphic [dot] com.

2 Likes
2

+1 to this. Just wanted to say that the case studies and chapters that he’s been putting out over the last year or so have been pretty outstanding. For some examples in the areas of the above listed courses, this case study on a subset of the Netflix data builds up a hierarchical ordinal model with hierarchical cutpoints, hierarchical movie affinities, and hierarchical customer affinities all within the same model. This case study implements pairwise comparison models on video game racing data and concludes with insights such as predictions of future races and race entrant rankings.

I came from a very methods-first background, where you learn how to use a bunch of different tools (regression, glm, etc) and then every problem you come across gets solved with one of the tools in the bag. This is a rather limiting approach. The thing that I like about Betancourt’s material is that it’s more of a problem-solving approach, where the goal is to model as best as possible the narratively generative story (i.e. how the data came to be). I have found this approach to be far more useful and powerful in practice. The modeling methods are taught with this in mind, rather than as simply another tool for the bag.