Effective hybrid feature selection using different bootstrap enhances cancers classification performance

Table 17 The proposed methods compared with the MIFS, CBF and FCBF methods

Datasets	Train Data %	Test Data %	Over-fitting Diff. %	Pre	Rec	F1-score	NO.F	F-Time (sec)	C-Time (sec)	AUC	Var.	ACC %
Mutual Information
KNN classifier
RNA gene	99.736	99.627	0.109	0.998	0.996	0.997	10,000	258.902	0.008	1.000	0.000036	99.627
DNA CNV	82.686	76.097	6.589	0.745	0.663	0.667	9000	180.314	0.011	0.854	0.000368	76.097
Parkinson’s disease	80.879	72.479	8.400	0.610	0.568	0.572	300	2.121	0.0001	0.624	0.002344	72.479
Dermatology diseases	97.966	97.267	0.699	0.975	0.969	0.969	25	0.351	0.002	0.963	0.000839	97.267
BreastEW	94.435	92.628	1.807	0.927	0.917	0.920	20	0.083	0.00002	0.958	0.001419	92.628
Correlation Based Feature
KNN classifier
RNA gene	99.867	99.748	0.119	0.999	0.997	0.998	900	2.600	0.003	1.000	0.000092	99.748
DNA CNV	52.831	49.073	3.758	0.447	0.402	0.369	750	1.850	0.003	0.669	0.000490	49.073
Parkinson’s disease	81.158	72.612	8.546	0.612	0.571	0.575	320	0.255	0.002	0.627	0.002308	72.612
Dermatology diseases	94.171	90.953	3.218	0.871	0.855	0.846	20	0.202	0.002	0.947	0.002243	90.953
BreastEW	94.747	92.976	1.771	0.931	0.920	0.924	17	0.105	0.002	0.961	0.000953	92.976
Fast Correlation Based Feature
KNN classifier
RNA gene	99.742	99.625	0.117	0.998	0.996	0.997	400	1.750	0.001	1.000	0.000131	99.625
DNA CNV	81.390	76.236	5.154	0.721	0.671	0.676	13	0.800	0.007	0.905	0.001131	76.236
Parkinson’s disease	82.657	73.270	9.387	73.271	0.585	0.587	16	1.500	0.002	0.675	0.001767	73.270
Dermatology diseases	97.936	97.005	0.931	0.970	0.967	0.966	14	0.101	0.002	0.961	0.001217	97.005
BreastEW	95.333	95.078	0.255	0.953	0.945	0.947	7	0.006	0.002	0.953	0.000261	95.078

ISSN: 1756-0381