A New Robust Classifier on Noise Domains: Bagging of Credal C4.5 Trees

Table 9

Friedman’s ranks about the accuracy of the algorithms when they are applied on data sets with different percentages of added noise.

Algorithm	Noise 0%	Noise 10%	Noise 20%	Noise 30%	Noise 40%
BA-C4.5-U	3.17	4.88	5.63	5.94	5.82
BA-CDT-U	4.52	4.42	4.85	4.81	4.89
BA-CC4.5-U	4.08	3.60	3.09	3.34	3.26
BA-C4.5	3.78	3.09	3.45	3.92	4.08
BA-CDT	5.43	4.27	3.07	2.25	2.34
BA-CC4.5	4.24	3.48	2.68	2.14	1.91
RF	2.78	4.26	5.23	5.60	5.70