DiscriMiner is an R package that provides functions for Discriminant Analysis and Classification purposes. Among the covered topics one can find methods for descriptive, geometric, linear, quadratic, PLS, as well as qualitative discriminant analyses.
Among my personal library of books about data analysis and multivariate statistics, I have a small collection of French textbooks. These are awesome books, very didactic, with clear explanations, good examples, and lots of graphics. They all share the underlying spirit of the French school known as Analyse des Données, something similar -but not identical- to the Exploratory Data Analysis approach of John Tuckey.
One drawback is that these books are written in French with, sadly, just some random limited editions in English. But the real problem (at least for me) was the lack of software availability for performing discriminant analysis in the way they are described in my books. Such was my frustration and annoyance, that I decided to create DiscriMiner in order to have software that allowed me to perform discriminant analysis, as well as getting the results, in the way they are explained in my french textbooks.
Stable version CRAN
# stable version
install.packages("DiscriMiner")Development version on github
# development version
library(devtools)
install_github('DiscriMiner', username='gastonstat')# load package
library(DiscriMiner)
# laod dataset
data(iris)
# apply geometric predictive discriminant analysis
geom_model = geoDA(iris[,1:4], iris$Species)
# show me the model
geom_modelMany thanks to Charles Determan Jr for his contributions to the function plsDA
Gaston Sanchez (gaston.stat at gmail.com)