A few weeks ago, I completed a graduate-level Bayesian Statistics course at Penn (STAT-927). Although I have been interested in Bayesian Statistics for quite a while now, this was the first formal course I have taken. Professionally, I developed some Bayesian multilevel regression models during my time at the Computational Memory Lab as part of the RAM project. However, in my current position, I have not typically leveraged fully Bayesian approaches.
Last week at work, I gave an overview of Bayesian Statistics during a lunch-and-learn style session. The intention was to have the material be accessible to anyone with some background in probability and at least a high-level understanding of classical statistics (e.g. at least some familiarity with p-values, hypothesis testing, and confidence intervals). The structure of the presentation largely follows the outline from the first couple of lectures from the course (albeit at a slightly higher level), but with the addition of a running example.
I had a lot of fun putting the slides together and wrestling with how to present the material to a mixed audience (data curators, data scientists, and software engineers). As putting a presentation together often does, it solidified my understanding in many places and pointed to some areas where it is still lacking. I’m hoping to delve into some additional techniques from the course in a future presentation (probably geared towards a smaller audience). In the meantime, here are the slides and code from the initial presentation.