Hello all!
Please pardon a little self-promotion. :)
My second, advanced Stan course is ready!
This course continues from where my first course ended (forum discussion), and took almost a year to develop the additional curricula, content, code, and videos that teach advanced Stan for Bayesian analysis in the context of sports examples.
As with the first course, this course format teaches actual Stan code instead of a higher level interface, videos where you watch me live-code each model with function-by-function, line-by-line explanations in a hosted RStudio session where you can practice alongside me.
My aim is to give you the closest thing to sitting with me in my office as we build models together.
As the topics are more advanced than the first course, the 136 videos span almost 20 hours: think three seasons of Ted Lasso but with fewer locker room dramas and more log-likelihoods.
While we’ll consider spatial data from the football (soccer) pitch, we’ll also model data-generating processes in sprinting performance, Formula One racing, baseball base running, umpire visual perception, team scoring, basketball three-point attempts, SailGP racing dynamics, golf putting mechanics, cycling training adaptations, tennis performance, and scouting evaluations.
Here’s a trailer of what we’ll build together:
Flexible modeling approaches:
Mixture models, over-dispersion, and zero-inflation
Splines, Gaussian processes, and Hilbert-space approximations
Physics-constrained models
Complex data structures:
Rating and ranking models, Plackett-Luce, Ordinal regression
Multi-level structures that propagate uncertainty
Correlation, trivariate reduction, copulas
Time and events:
Autoregressive processes
Survival analyses: continuous and discrete time
Differential and difference equations
Computational optimization:
QR reparameterization, sufficient statistics
Parallelization and other performance considerations
Be curious, not judgmental. — Ted Lasso
That’s the spirit of this course: curiosity-driven, principled modeling rooted in real-world sports data.
If you or someone you know are ready to go deeper—to expand your modeling toolkit, wrestle with richer data structures, and sharpen your command of Stan—I hope you’ll join me for this next step.
Let’s keep building. Let’s keep learning.
As always, feel free to ask any questions about the course here or via DM.
Believe.