Boosting Diversity in Regression Ensembles.

Bodi

This R package implements the Bodi algorithm. The goal of Bodi (BOosting DIversity) is to construct a sequence of learners actively promoting diversity among them. Then, increasing diversity ensures the reduction of the mean-square-error producing a strong(er) ensemble learner.

Installation

You can install the released version of Bodi from CRAN with (soon):

install.packages("Bodi")

or from Gitlab (need to make repo public or use a PAT)

remotes::install_gitlab("yannig/bodi")

Example

Bodi has a main function called boosting_diversity. This basic example shows how to obtain the Bodi fit of a simple dataset :

library(Bodi)
all <- na.omit(airquality)
smp <- sample(nrow(all), .8 * nrow(all))
boosting_diversity("Ozone", "Solar.R+Wind+Temp+Month+Day", 
                   data0 = all[smp, ], data1 = all[-smp, ])

See the documentation for further details.

Acknowledgements

This research development benefited from the support of the «FMJH Research Initiative Data Science for Industry», and from the support to this program from EDF.