Articles
Introduction
This vignette provides a broad introduction to the various facilities in the package. A good place to start if you are new to the package.
Model Families
If you wish to use model families ala stats::glm, this vignette is for you. The family also includes Cox models for survival, which are discussed in the next vignette below.
Survival Models
Penalized Cox regression models can now be fit in glmnet using both right censored and (start, stop] data. This vignette leads you through the process with several examples.
Relaxed Fits
The idea of the relaxed lasso is to take a glmnet fitted object, and then for each lambda, refit the variables in the active set without any penalization. For details, see below.
History
A scholarly history of glmnet across its major versions, with attribution to contributors and links to published references.
Technical Details for Cox Models
Cox models come with some interesting technical details, in particular the computation of the gradient, Hessian and deviance. These change when start/stop is used rather than right censored, when there are ties in death times, and when there are strata. We had difficulty finding a good reference and efficient computational algorithms for these calculations, so produced this writeup.