A Novel Robust Fuzzy Rough Set Model for Feature Selection

Table 4

Comparison of classification performance of different FRS models on LSVM in original and noisy data.

Datasets	Noise level (%)	FRS	β-PFRS	K-trimmed FRS	K-means FRS	K-median FRS	SFRS	RS-FRS
Glass	0	0.548 ± 0.122	0.571 ± 0.126	0.570 ± 0.128	0.581 ± 0.128	0.575 ± 0.128	0.582 ± 0.122	0.548 ± 0.122
	5	0.542 ± 0.034	0.569 ± 0.021	0.567 ± 0.007	0.575 ± 0.033	0.574 ± 0.051	0.580 ± 0.085	0.547 ± 0.090
	10	0.539 ± 0.028	0.564 ± 0.015	0.565 ± 0.027	0.571 ± 0.019	0.572 ± 0.029	0.577 ± 0.060	0.545 ± 0.110
Wine	0	0.938 ± 0.055	0.950 ± 0.072	0.950 ± 0.072	0.938 ± 0.055	0.938 ± 0.055	0.948 ± 0.055	0.949 ± 0.068
	5	0.892 ± 0.062	0.945 ± 0.055	0.936 ± 0.079	0.937 ± 0.039	0.935 ± 0.050	0.946 ± 0.034	0.947 ± 0.101
	10	0.890 ± 0.064	0.940 ± 0.027	0.935 ± 0.070	0.933 ± 0.029	0.935 ± 0.056	0.941 ± 0.082	0.942 ± 0.069
Heart	0	0.770 ± 0.087	0.830 ± 0.058	0.778 ± 0.070	0.767 ± 0.080	0.767 ± 0.080	0.770 ± 0.087	0.770 ± 0.087
	5	0.767 ± 0.080	0.781 ± 0.044	0.772 ± 0.086	0.761 ± 0.051	0.760 ± 0.039	0.768 ± 0.066	0.766 ± 0.043
	10	0.753 ± 0.043	0.754 ± 0.055	0.752 ± 0.065	0.741 ± 0.012	0.752 ± 0.025	0.755 ± 0.038	0.765 ± 0.071
Segment	0	0.903 ± 0.055	0.904 ± 0.054	0.904 ± 0.053	0.905 ± 0.052	0.904 ± 0.053	0.905 ± 0.052	0.905 ± 0.052
	5	0.894 ± 0.057	0.896 ± 0.018	0.901 ± 0.087	0.895 ± 0.010	0.895 ± 0.071	0.899 ± 0.025	0.903 ± 0.044
	10	0.874 ± 0.020	0.876 ± 0.042	0.875 ± 0.059	0.877 ± 0.076	0.878 ± 0.093	0.894 ± 0.101	0.897 ± 0.069
Hepatitis	0	0.815 ± 0.069	0.815 ± 0.080	0.815 ± 0.080	0.815 ± 0.080	0.815 ± 0.080	0.845 ± 0.080	0.845 ± 0.080
	5	0.805 ± 0.057	0.811 ± 0.036	0.809 ± 0.061	0.814 ± 0.027	0.813 ± 0.054	0.832 ± 0.099	0.834 ± 0.096
	10	0.793 ± 0.091	0.801 ± 0.061	0.815 ± 0.045	0.812 ± 0.052	0.811 ± 0.059	0.833 ± 0.067	0.833 ± 0.013
ICU	0	0.926 ± 0.023	0.926 ± 0.023	0.926 ± 0.023	0.926 ± 0.023	0.926 ± 0.023	0.926 ± 0.023	0.926 ± 0.023
	5	0.910 ± 0.047	0.920 ± 0.018	0.920 ± 0.023	0.923 ± 0.033	0.919 ± 0.039	0.921 ± 0.071	0.924 ± 0.104
	10	0.899 ± 0.013	0.911 ± 0.047	0.910 ± 0.043	0.903 ± 0.022	0.906 ± 0.019	0.919 ± 0.045	0.922 ± 0.055
German	0	0.735 ± 0.054	0.739 ± 0.057	0.739 ± 0.057	0.742 ± 0.055	0.742 ± 0.055	0.732 ± 0.054	0.735 ± 0.054
	5	0.720 ± 0.055	0.722 ± 0.104	0.725 ± 0.046	0.725 ± 0.115	0.729 ± 0.048	0.727 ± 0.015	0.731 ± 0.080
	10	0.701 ± 0.047	0.714 ± 0.054	0.711 ± 0.048	0.702 ± 0.044	0.708 ± 0.041	0.719 ± 0.039	0.725 ± 0.011
Soy	0	0.765 ± 0.024	0.775 ± 0.024	0.765 ± 0.024	0.765 ± 0.024	0.765 ± 0.024	0.765 ± 0.063	0.775 ± 0.024
	5	0.705 ± 0.032	0.740 ± 0.025	0.736 ± 0.039	0.740 ± 0.033	0.738 ± 0.074	0.739 ± 0.017	0.763 ± 0.117
	10	0.700 ± 0.047	0.731 ± 0.063	0.724 ± 0.032	0.722 ± 0.018	0.721 ± 0.058	0.735 ± 0.023	0.753 ± 0.027
Horse	0	0.832 ± 0.052	0.829 ± 0.052	0.842 ± 0.057	0.842 ± 0.057	0.834 ± 0.053	0.839 ± 0.052	0.832 ± 0.052
	5	0.826 ± 0.059	0.828 ± 0.069	0.834 ± 0.053	0.835 ± 0.044	0.831 ± 0.034	0.834 ± 0.073	0.831 ± 0.054
	10	0.807 ± 0.044	0.824 ± 0.068	0.832 ± 0.057	0.823 ± 0.050	0.821 ± 0.059	0.829 ± 0.026	0.830 ± 0.080
WDBC	0	0.935 ± 0.035	0.943 ± 0.020	0.935 ± 0.046	0.933 ± 0.047	0.939 ± 0.047	0.935 ± 0.035	0.935 ± 0.035
	5	0.930 ± 0.042	0.942 ± 0.025	0.933 ± 0.032	0.931 ± 0.035	0.933 ± 0.039	0.934 ± 0.059	0.934 ± 0.100
	10	0.923 ± 0.040	0.933 ± 0.021	0.932 ± 0.044	0.930 ± 0.045	0.926 ± 0.047	0.931 ± 0.061	0.933 ± 0.017
WPBC	0	0.763 ± 0.030	0.763 ± 0.030	0.763 ± 0.030	0.763 ± 0.030	0.763 ± 0.030	0.763 ± 0.030	0.763 ± 0.030
	5	0.711 ± 0.027	0.716 ± 0.015	0.721 ± 0.052	0.734 ± 0.024	0.733 ± 0.068	0.762 ± 0.080	0.734 ± 0.087
	10	0.705 ± 0.082	0.714 ± 0.109	0.720 ± 0.076	0.724 ± 0.064	0.730 ± 0.099	0.759 ± 0.055	0.730 ± 0.020
Sonar	0	0.712 ± 0.072	0.725 ± 0.071	0.716 ± 0.089	0.721 ± 0.110	0.721 ± 0.110	0.722 ± 0.072	0.733 ± 0.055
	5	0.711 ± 0.069	0.719 ± 0.101	0.714 ± 0.076	0.717 ± 0.119	0.716 ± 0.119	0.715 ± 0.068	0.728 ± 0.038
	10	0.673 ± 0.026	0.716 ± 0.012	0.712 ± 0.034	0.711 ± 0.097	0.713 ± 0.025	0.713 ± 0.024	0.718 ± 0.073
Average		0.786 ± 0.051	0.801 ± 0.049	0.799 ± 0.055	0.798 ± 0.051	0.798 ± 0.057	0.805 ± 0.057	0.804 ± 0.064