A Novel Feature Selection Method for Classification of Medical Data Using Filters, Wrappers, and Embedded Approaches

Algorithm 1

The algorithm of proposed feature selection method is given as follows:

	Input F: Original feature set
	N: Size of population
	D: Dimensions of feature
	O: Optimal feature subset
	For each F = 1-N Do:
	While (each feature selection method not evaluated)
	Evaluate each feature using feature selection methods
	Calculate weight of each feature
	Select top features based on weight
	Evaluate each feature selection method on selected features
	Calculate Accuracy, Sensitivity, Specificity, F-Measure
	Select Top 3 methods based on Accuracy
	If feature selection method Accuracy >80
	Then select feature selection method
	Select features subset D of each feature selection method
	Perform intersection on selected features subsets D
	Return O: Optimal feature subset
	Else ignore
	End While
	End For
	Output O: Optimal feature subset