This is the GitHub repository for the GeoSci Seminar in Winter 2023 led by Mark Webster
Here are a few resources I (Jon) find indispensable when working with Bayesian multilevel models:
- brms R package website
- brms is an all-in-one Stan interface for Bayesian regression modeling that uses a simple syntax and is incredibly powerful and user friendly. Vignettes, Functions, etc. can be found here.
- brms is an all-in-one Stan interface for Bayesian regression modeling that uses a simple syntax and is incredibly powerful and user friendly. Vignettes, Functions, etc. can be found here.
- Statistical Rethinking v.2 with
brms
,ggplot
, and thetidyverse
- This is undoubtedly the most useful tool for my forrays into modeling. @SolomonKurz goes chapter by chapter though Statistical Rethinking v.2 and works out all of the models in
brms
, generates all of the figures inggplot
, and does all the data wrangling in thetidyverse
. (Note: if you are not wedded to a data formatting methodology like base R orpandas
, check out the tidyverse sooner rather than later.) Plus there are some tidbits throughout the text that really take it to the next level, such as using great color palettes, using patchwork for plot layouts, and excellent bonus sections like this one on different ways to model categorical predictor variables
- This is undoubtedly the most useful tool for my forrays into modeling. @SolomonKurz goes chapter by chapter though Statistical Rethinking v.2 and works out all of the models in
-
Probability Distribution Applet
- What does an Exponential Distribution with a
$\lambda$ of 3 look like? How thick are the tails on a Student-$T$ with a$\nu$ parameter of 4? Do I use a Poisson on Negative-Binomial Distribution for these count data? This applet makes visualizing continuous distributions really easy. There is a mobile app too.
- What does an Exponential Distribution with a
- tidybayes
- An excellent tidy way to extract draws, predictions, and plot uncertainty for Bayesian models.
- An excellent tidy way to extract draws, predictions, and plot uncertainty for Bayesian models.
- Stan Discourse Forum
- First place to check for any questions regarding Stan, brms, and Rethinking. Topics range from OS issues ("R not compiling my models properly") to conceptual topics to new features or methods.