This post is by Aki
Mine Dogucu and Jingchen Hu arxived in September a paper The Current State of Undergraduate Bayesian Education and Recommendations for the Future with an abstract
With the advances in tools and the rise of popularity, Bayesian statistics is becoming more important for undergraduates. In this study, we surveyed whether an undergraduate Bayesian course is offered or not in our sample of 152 high-ranking research universities and liberal arts colleges. For each identified Bayesian course, we examined how it fits into the institution’s undergraduate curricula, such as majors and prerequisites. Through a series of course syllabi analyses, we explored the topics covered and their popularity in these courses, the adopted teaching and learning tools, such as software. This paper presents our findings on the current practices of Bayesian education at the undergraduate level. Based on our findings, we provide recommendations for programs that may consider offering Bayesian education to their students.
It’s not mentioned in the abstract, but based on the references “U.S. News (2021a), ‘2021 Best national university rankings’” and “U.S. News (2021b), ‘National liberal arts colleges’” the current state is only about US research universities and liberal arts colleges.
Anyway, I highly recommend checking out their paper as in addition to the survey of the state, they make recommendations for the future:
- Expand the access to Bayesian courses
- Make Bayesian courses a part of the majors
- Balance statistics with computing
- Use a variety of assessments
It’s easy for me to agree with these, as I’ve been doing these for years: 1) I’m making my course available for everyone, 2) my course is optional in Data science BSc major (and some other majors), and required in Machine learning / AI MSc major at Aalto University (the course is quite advanced, and we’re missing an intermediate course at BSc level, so that’s why it’s only optional at BSc level), 4) I use a variety of assessments. Especially, I like the subpoints in the recommendation 3:
- Introduce simulation-based learning early in the course.
- Encourage students to write self-coded MCMC algorithms for relatively simple multiparameter models
- If the course puts equal emphasis on computing and modeling, consider adopting one of the popular probabilistic programming languages for Bayesian model estimation through MCMC (e.g., JAGS and Stan).
- If the course has a slightly stronger emphasis on modeling over computing, consider introducing one of the wrapper packages for Stan for its simpler posterior summary procedure (e.g., rstanarm and brms).
I have been including all these in my course (except JAGS), so I’m already in the future :D
I’m also happy to see that the most commonly required or recommended book is still 8 year old “Bayesian Data Analysis”.
Dogucu and Hu end their paper with
Last but not least, we invite current and future Bayesian educators to join the undergraduate Bayesian education network, an online community that fosters discussions of undergraduate Bayesian education.