Fibonacci and k-Subsecting Recursive Feature Elimination

Source codes and experimental scripts for Fibonacci and k-Subsecting Recursive Elimination feature selection methods. Benchmark data sets taken from the scikit-feature selection repository (http://featureselection.asu.edu/).

Algorithms described in "Fibonacci and k-Subsecting Recursive Feature Elimination". Detailed experimental results available at: http://www.cs.put.poznan.pl/dbrzezinski/software/SRFE.html

Installation

Dependencies

The srfe package was tested to work under Python 2.7 and Python 3.5. The main packages used are:

scipy
numpy
pandas
scikit-learn
lighgbm

A detailed list of packages with versions used during experiments are in requirements.txt.

Installation

Download or clone the repository and run the setup.py file. Use the following commands to get a copy from GitHub and install all dependencies:

  git clone https://github.com/dabrze/subsecting_rfe.git
  cd subsecting_rfe
  pip install .