Table 4.

Utilized the set of fine-tuned parameters for employed baseline classifiers in different feature sets by using GridsearchCV.

Feature set	RF	DT	KNN
ALL Features	n_estimator = 10, random_state = 10, max_depth = 7, max_features = ‘sqrt’	max_depth = 5, random_state = 5,	n_neighbors = 10, leaf_size = 25,
		criterion = ‘entropy’,	metric = ‘minkowski’,
		splitter = ‘best’	algorithm = ‘brute’
UVS Features	n_estimator = 8, random_state = 15, max_depth = 5, max_features = ‘sqrt’	max_depth = 8, random_state = 15,	n_neighbors = 15, leaf_size = 30,
		criterion = ‘gini’,	metric = ‘minkowski’,
		splitter = ‘best’	algorithm = ‘auto’
IGS Features	n_estimator = 7, random_state = 10, max_depth = 10, max_features = ‘sqrt’	max_depth = 5, random_state = 20,	n_neighbors = 5, leaf_size = 30,
		criterion = ‘gini’,	metric = ‘minkowski’,
		splitter = ‘best’	algorithm = ‘auto’