Statistical-Rethinking-Notes

This is my collection of notes on Statistical Rethinking: A Bayesian Course with Examples in R and Stan. These are offered without assurances, besides the guarantee of mistakes and typos.

How to use

There is a folder for each chapter and in each folder there are files for notes and exercises. There are two types of files, .Rmd and .md. To see the results in formatted html open the .md files on github. Some code or warnings may be suppressed. For the complete code, open the .Rmd files in RStudio.

Other sets of notes

Self-study is difficult without feedback. I used the following sets of notes to check my understanding of the material.

https://github.com/cavaunpeu/statistical-rethinking

https://github.com/rlabbe/statistical-rethinking

http://scs.math.yorku.ca/index.php/SCS_Reads_2016-2017

Why my notes are different

The book follows a traditional r-base style along with some added utility and Stan functions from the rethinking package. The other notes mostly conform to this style.

As my notes progress I explore new packages while trying to verify the results using the expected methods. Following along with the book is a good way for me to build confidence and experience with new tools.

Here are some of the packages that I incorporate:

1. tidyverse: This is a collection of packages that attempt to make a consistent data science workflow. In particular, I explore

   a. ggplot2 for visualization.

   b. dplyr for tidying data frames.

   c. modelr for data_grid to easily generate new data for models.

2. tidybayes to extend the tidy philosophy to Bayesian tools.

3. brms as an alternative interface to Stan. This package allows you to construst full Bayesian models in Stan using the standard formula syntax, while offering some additional functionality to rethinking::map2stan. The package comes with a suite of helper functions that automate many of the common processes in the book, such as building the inverse-link function and calculating percentiles. For a comparsion to the rethinking package, check out the brms overview.

Contact and Discussion

If you have any questions or suggestions you can reach me anytime at [email protected], or you can leave an issue on this repo.

For more general Rethinking questions Lotty Brand recently started a Rethinking chat room on slack. The author Richard McElreath has also joined the forum.