Algorithm 1.

Pseudo Code for Recursive Feature Elimination (RFE) Algorithm

Input:

Training set: T

Feature set: F = {f1, f2, …, fp}

Ranking method: M(T, F)

Desired feature number: q

Number of feature to eliminate in each step: k

Output:

Final ranking feature set: R = {fr1, fr2, …, frp}

Final selected feature set: F = {f1, f2, …, fq}

1 Initialization;

2 Steps: S = (p-q)/k;

3 for i = 1 → S do

$\begin{array}{cc}\begin{array}{l}4\\ 5\\ 6\\ 7\end{array}& |\begin{array}{l}\text{ Rank set }F\text{ according to }M\text{(}T,F\text{);}\\ L_k\leftarrow \text{Last ranked }k\text{ features in }F;\\ R[p-i\ast k+1:p-(i-1)\ast k]\leftarrow L_k;\\ F\leftarrow F-L_k;\end{array}\end{array}$

8 end