Jx-FFST : A Filter Feature Selection Toolbox

---

"Toward Talent Scientist: Sharing and Learning Together" --- Jingwei Too

---

Introduction

• This toolbox contains 4 filter feature selection methods
• The A_Main file shows the example of how to use these filter feature selection on the benchmark dataset

Usage

The main function jffs is employed to perform feature selection. You may switch the method by changing the 'rf' to other abbreviations

• If you wish to use Relief-F ( RF ) then you may write

FS = jffs('rf',feat,label,opts);

Input

• feat : feature vector matrix ( Instance x Features )
• label : label matrix ( Instance x 1 )
• opts : parameter settings
  • Nf : number of selected features ( must be set for all methods )

Output

• Acc : accuracy of validation model
• FS : feature selection model ( It contains several results )
  • sf : index of selected features
  • ff : selected features
  • nf : number of selected features
  • s : weight / score
  • t : computational time (s)

Example : Relief-F ( RF )

% Parameters
opts.K  = 3;     % number of nearest neighbors
opts.Nf = 10;    % select 10 features

% Load dataset
load ionosphere.mat; 

% Perform feature selection 
FS     = jffs('rf',feat,label,opts);

% Define index of selected features
sf_idx = FS.sf;

% Accuracy  
kfold  = 5;
Acc    = mSVM(feat(:,sf_idx),label,kfold);

Requirement

• MATLAB 2016 or above
• Statistics and Machine Learning Toolbox

List of available filter feature selection methods

• The extra parameter represent the parameter other than number of selected features
• Note : Relief-F contains extra parameter to be adjusted. If you do not set the parameter then the algorithm will be executed using default setting
• You can use opts to set the parameter
  • K : number of nearest neighbors

No.	Abbreviation	Name	Extra Parameter
4	'nca'	Neighborhood Component Analysis	-
3	'rf'	Relief-F	opts.K = 5
2	'tv'	Term Variance	-
1	'pcc'	Pearson Correlation Coefficient	-