README.md

In matpato/EEFR: Enhanced Ensemble Feature Ranking (EEFR)

EEFRanking (Enhanced Ensemble Feature Ranking)

EEFRanking R package implements EEFR feature selection algorithm as an enhancement of the former EFR. This project is part of my Master's Degree on Computer Science and Engineering at Instituto Superior de Engenharia de Lisboa (ISEL).

The aim of EEFR is to make Feature Ranking and Feature Selection tasks easier for Data Mining (DM) developers in R. On a DM process, EEFR can be used as feature selection pre-process, before classification.

Package installation from GitHub:

library("devtools") install_github('https://github.com/matpato/EEFR.git', INSTALL_opts=c("--no-multiarch"))

Package usage:

library(EEFRanking) features <- ensemble.features.ranking(dataset) dataset <- dataset[,c(features, 'class')]

Package documentation:

The feature selection is executable by ensemble.features.ranking(dataset, ...). A list of optional parameters is available, as specified on the documentation. As output, it returns a list with the name of the features sorted in inverse order of relevance. The returned list can include all the sorted features or a subset of the first most relevant ones.

EEFRanking documentation is accessible by: help(package="EEFRanking"). The R package includes a demo example, executable by: example("ensemble.features.ranking"). The example uses an artificial dataset included in R package.

Package EEFRanking is accessible from https://github.com/matpato/EEFR.git repository The package pEEFRanking (parallel version) is accessible from https://github.com/matpato/EEFR-parallel.git repository

matpato/EEFR documentation built on Dec. 21, 2021, 2:53 p.m.