featscreen

featscreen is an R package providing ready-to-use functions for straightforward feature screening in both supervised and unsupervised settings.

featscreen is crafted with the overarching goal of simplifying and unifying commonly used screening techniques into a single, accessible R package.

Explore the variety of supervised and unsupervised screening techniques available in featscreen to tailor your feature selection process according to your analysis needs.

The following table provides an overview of the supervised methods currently supported in the package:

category	name	numerical	categorical	survival
correlation	Kendall’s rank correlation coefficient t-test	x
correlation	Pearson’s product moment correlation coefficient t-test	x
correlation	Spearman’s rank correlation coefficient t-test	x
two-groups	two-sample Student’s pooled t-test		x
two-groups	paired two-sample Student’s t-test		x
two-groups	two-sample t-test with the Welch modification to the degrees of freedom		x
two-groups	paired two-sample Wilcoxon signed-rank test		x
two-groups	two-sample Mann-Whitney U-test		x
multi-groups	one-way analysis of variance F-test		x
multi-groups	one-way analysis of variance F-test with Welch correction		x
multi-groups	Pearson’s χ²-test		x
multi-groups	Kruskal-Wallis H-test		x
survival	Cox PH regression coefficient z-test			x
biostatistic	empirical Bayes moderated F-test	x	x
biostatistic	empirical Bayes moderated t-test	x	x
biostatistic	significant analysis of microarrays permutation test	x	x	x

Continuing our exploration of feature screening capabilities in featscreen, the following table offers insights into the unsupervised screening methods integrated into the package:

name	numerical	categorical	multiresponse	survival
above-median frequency ratio	x
above-minimum frequency ratio	x
median value	x
missing value ratio	x	x	x	x
variability	x

Installation

You can install latest development version from GitHub (requires devtools package):

if (!require("devtools")) {
  install.packages("devtools")
}

devtools::install_github(
  repo = "alebarberis/featscreen", 
  dependencies = TRUE, 
  build_vignettes = FALSE
)

Getting started

If you are just getting started with featscreen, we recommend looking at the Getting Started section of the site.

Credits

featscreen was conceived, developed, and is maintained by Alessandro Barberis (@alebarberis).

Acknowledgements

I would like to express my sincere gratitude to Prostate Cancer UK for their generous funding, which made it possible for me to develop the first version of this package.