GrpClassifierEC: a novel classification approach based on the ensemble clustering space

Table 6 GrpClassifierEC: EC classifier results with a k value of 30 compared to Random forest applied on the EC samples and results for regular classifiers applied on the original data

Data/performance	Data info			EC classifier GrpClassifierEC				Accuracy difference
Data/performance	#Sample	#EC_Samples	ratio	Sensitivity	Specificity	F-measure	Accuracy	EC-RF	RF	DTT	KNN
Aves vs Embryophyta	1068	513	48%	0.86	0.94	0.85	0.92	-0.01	-0.03	0.02	-0.01
Cercopithecidae vs Malvaceae	894	449	50%	0.94	0.92	0.94	0.94	0.04	0.01	0.06	0.03
Embryophyta vs Laurasiatheria	953	493	52%	0.94	0.83	0.94	0.91	0.04	0.00	0.06	0.03
Fabaceae vs Nematoda	2642	536	20%	0.78	0.88	0.79	0.84	-0.01	-0.05	0.01	-0.04
Hexapoda vs Aves	2840	1647	58%	0.76	0.92	0.78	0.88	0.05	-0.01	0.07	0.06
Laurasiatheria vs Brassicaceae	1209	406	34%	0.89	0.88	0.89	0.88	0.00	-0.04	0.00	-0.03
Malvaceae vs Fabaceae	1401	451	32%	0.55	0.80	0.53	0.73	0.07	-0.04	0.06	0.03
brassicaceae vs Hexapoda	2584	542	21%	0.77	0.95	0.78	0.91	-0.01	-0.03	0.01	-0.02
Hominidae vs Cercopithecidae	1829	786	43%	0.61	0.87	0.63	0.80	0.10	0.04	0.14	0.09
Monocotyledons vs HomoSapiens	2625	855	33%	0.86	0.87	0.86	0.87	0.04	-0.03	0.03	-0.01
Average			39%	80%	89%	80%	87%	3%	-2%	5%	1%

K is number of clusters. The section “Accuracy Difference” is EC Classifier-ACC of the other classifier. A positive value indicates that the EC classifier is better than the other corresponding classifiers. EC-RF is a random forest applied on the EC data, RF is a random forest applied on the original data. DTT is a decisionTrees while KNN is K- Nearest Neighbors applied on the original data

ISSN: 1748-7188