Algorithm 1: Hybrid filter and differential evolution-based feature selection
	Input: F: Original feature set, and P: Size of population Output: SF: The optimal selected features
	Begin
1	Compute scores of the features using Filter methods: Information gain (IG), information gain ratio (IGR), correlation (CR), Gini index (GIND), Relief (RELIEF) and Chi-squared (CHSQR)
2	Do ranking of all features in F based on the scores computed by Filter methods
3	Reduce dimension of training data by selecting only the top 5% of ranked features
4	Set Crossover rate, maximum number of generations Max_t, the generation counter t = 0
5	Generate and initialize P individuals of population (feature subset solutions)
6	While t < Max_t and Stopping criterion is not satisfied Do
7	t = t + 1
8	Compute the fitness of individuals using misclassification rate
9	Best individual = Evaluate the fitness of individuals
10	For each individual $X_i$ Do
11	Choose three individuals $X_{r_1}$, $X_{r_2}$, $X_{r_3}$ randomly   where $r1\ne r2\ne r3\ne i$
12	Generate a mutant solution $V_i$ using Equations (9) and (10)
13	Generate a trial vector $U_i$ using Equation (11)
14	Evaluate fitness values of $X_i$ and $U_i$
15	If fitness of $U_i$ is better than fitness of $X_i$
16	$X_i$ is replaced with $U_i$ in the population and then         $U_i$ used for the next generation
17	Else
18	$X_i$ is kept for next generation
19	End if
20	End For
21	End While
22	Extract the optimal selected features from the individual with the best fitness value
23	Return the optimal features SF
24	End Algorithm