Feature quantization for parsimonious and interpretable models

Table of Contents

• Documentation
• Installation instructions
• Open an issue

glmdisc

The goal of glmdisc is to discretize continuous features, merge factor levels, and sparsely introduce interactions in logistic regression models.

This is the implementation of glmdisc-SEM as described in Formalization and study of statistical problems in Credit Scoring, Ehrhardt A. (see manuscript or web article)

Installation

You can install the development version of glmdisc from Github with:

# install.packages("devtools")
devtools::install_github("adimajo/glmdisc", build_vignettes = TRUE)

Or alternatively directly from CRAN:

install.packages("glmdisc")

Documentation

Through R(Studio)

Getting help

The help pages of this package can be reached by typing help(glmdisc) (or ?glmdisc) once the package is installed and loaded.

Vignette

For further instructions, go to vignettes > glmdisc.Rmd or go through the vignette directly in RStudio by typing vignette('glmdisc') once the package is installed and loaded.

Online

Both the help pages and the vignette are available online as a Github page built with pkgdown.