A Rough Hypercuboid Approach for Feature Selection in Approximation Spaces
Citations
181 citations
Cites methods from "A Rough Hypercuboid Approach for Fe..."
...The classical rough set model, introduced by Pawlak [35], has been successfully used as a feature selection tool [22], [24], [28], [39], [46]....
[...]
153 citations
80 citations
69 citations
60 citations
References
40,147 citations
"A Rough Hypercuboid Approach for Fe..." refers background in this paper
...However, the proposed method provides highest classification accuracy of the SVM at 0:0 0:4 for Satimage data, and 0:2 0:4, ¼ 0:6 and 0.7 for Leukemia data....
[...]
...Hence, the proposed indices such as , ?, and can be used to act as the objective function of the feature selection algorithm in approximation spaces as they reflect good quantitative measures like existing SVM and C4.5....
[...]
...The method due to Chen and Wasikowski [35] for binary class data sets achieves 63.16, 81.48, 80.54, 68.75, and 85.29 percent accuracy using the SVM, and 52.63, 92.02, 80.53, 61.25, and 88.24 percent accuracy using the C4.5 on Breast Cancer, Ionosphere, Lung, DLBCLNIH, Prostate Cancer data, respectively....
[...]
...On the other hand, the MRMS criterion attains higher classification accuracy of the SVM and HR value only for Satimage data, while the MDMS criterion achieves higher classification accuracy of both SVM and C4.5 for Satimage data, and lower value of S index for Breast Cancer data....
[...]
...Subsequent discussions analyze the results with respect to various proposed quantitative indices such as , ?, and , and the classification accuracy of both SVM and C4.5....
[...]
17,177 citations
14,948 citations
14,509 citations
"A Rough Hypercuboid Approach for Fe..." refers background in this paper
...It is to search a set of features that approximates Max-Dependency criterion with the mean value of all dependency values between an individual feature and the target class label [26], [27], [28]....
[...]