Whilst I’m preparing an annotated set of scripts for the Statistical Learning chapter of DPER5, you can access the updated R code used for the chapter via my dper5 github repository. The following scripts cover the examples from the chapter
- Univariate trees
- Multivariate regression trees
- Multivariate adaptive regression splines (MARS)
- Principal curves
- Self-organising maps (SOMs)
- Shrinkage methods in regression (ridge, LASSO, elastic net)
The boosted tree example is not yet posted, partly because I need to get permission to upload the EDDI diatom data used in the example.
The principal curves script requires the development version of my analogue package. Unfortunately, R-Forge is not building the package at the moment (there is an issue with the installation of the rgl package on their build servers). If you are using Linux then you can grab the source code and compile yourself. For MS Windows users, a version that works for the script, has been built using Uwe Ligge’s WinBuilder service. Download the zipfile, which was built using R 3.0.2-patched.