An Amalgamated Approach to Bilevel Feature Selection Techniques Utilizing Soft Computing Methods for Classifying Colon Cancer

Table 6

Average performance analysis of classifiers in terms of parameters with average to six optimization techniques for different gene selection methods using 30-60-90 selected genes.

Classifiers	Gene selection	Parameters (%)
		Accuracy	Perfect Classification	Performance Index	GDR
RF	30	83.59433	67.18491	45.30313	66.88702
	60	87.60384	75.02862	62.93849	74.75641
	90	82.42909	64.85208	40.91995	64.85275
Adaboost	30	83.87583	67.49332	46.09174	65.12554
	60	87.02927	74.44755	59.07174	67.71936
	90	83.56061	67.12004	45.80861	59.94513
LR	30	85.74922	71.49516	54.40771	59.299
	60	84.01784	68.03036	44.20435	46.31734
	90	88.0159	76.02367	65.72766	67.39558
DT	30	89.02009	78.04018	64.62035	71.93313
	60	88.21978	76.84355	64.60034	71.21466
	90	87.8683	75.7366	61.76893	75.73805
QDA	30	87.4349	74.86979	58.91599	74.53245
	60	88.93083	78.09815	67.8271	77.85126
	90	91.22996	82.36952	72.44417	82.37038