We’re a little over halfway through the stats class offered online by Stanford, and it’s rocking my world. So, so good. It helps that I have projects at work that involve all of these concepts.

The other day I had a look at the source code for the glmnet package, because I had a question on how the lambda coefficients controlling regularization for ridge regression and the Lasso were generated. I downloaded it from CRAN and poked around the file structure. It was neat to see how the R code called the Fortran, and it gives me a much better idea of how I would go to write a package.