The binomialRF package is an R package that provides a feature selection algorithm to be used in randomForest classifiers. Treating each tree as a quasi binomial stochastic process in a random forest, binomialRF determines a feature’s importance by how often they are selected in practice vs. as expected by random chance. Given that trees are co-dependent as they subsample the same data, a theoretical adjustment is made using a generalization of the binomial distribution that adds a parameter to model correlation/association between trials.
The binomialRF R package is on CRAN, and you can install as follows:
The CRAN version will always be the most stable release.
To install experimental updates from the binomialRF , install it from GitHub directly, follow the code instructions below!
install.packages("devtools") # The following dependencies might need to be installed # manually if they're not installed by devtools. install.packages(c("ggplot2", "randomForest", "data.table","rlist", "correlbinom")) devtools::install_github("SamirRachidZaim/binomialRF") library(binomialRF)
These GitHub updates and features are experimental and will not be available in the CRAN version until the next, stable release is pushed.
The main manuscript is included as a preprint in bioRxiv: https://doi.org/10.1101/681973, and has also been submitted for consideration at Frontiers in Genetics.
Any scripts or data that you put into this service are public.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.