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

Table 4

Performance analysis of classifiers in terms of PI with six optimization techniques for different gene selection methods using 30-60-90 selected genes.

Classifiers	Gene selection	Optimization techniques
		Invasive Weed Optimization	Teaching Learning-Based Optimization	League Championship Optimization	Beetle Antennae Search Optimization	Crow Search Optimization	Fruit fly Optimization
RF	30	60.105	85.7	78.4275	8.64875	8.64875	30.28875
	60	58.02563	86.37	84.66125	6.27625	76.62938	65.66844
	90	6.511875	43.225	78.93	7.69	25.80313	83.35969
Adaboost	30	54.54	59.85	78.465	5.805	4.39125	73.49922
	60	86.37	85.18063	90.78	18.14688	13.08297	60.87
	90	81.4325	48.1615	60.105	9.367813	4.8625	70.92234
LR	30	37.86063	77.17125	87.71	16.64625	20.29063	86.7675
	60	12.48375	63.30375	77.925	9.487656	4.744688	97.28125
	90	74.32875	78.93	13.86195	84.315	59.34	83.59028
DT	30	94.49	15.18023	18.36125	91.18	81.4325	87.07812
	60	90.78	76.62938	23.8125	18.52203	90.78	87.07812
	90	31.4425	24.21063	88.38	85.7	95.07	45.81047
QDA	30	10.32656	15.36	85.18063	62.3175	84.66125	95.65
	60	72.795	80.72125	87.04	15.18023	59.64609	91.58
	90	98.935	81.4325	81.4325	8.169375	74.04878	90.64688
Average		58.02848	61.42841	69.00484	29.83018	46.89546	76.67274