Fig 1.
uEFS methodology.
Fig 2.
UFS algorithm [19].
Fig 3.
TVS algorithm.
Table 1.
Predictive accuracy (in %age) of classifiers using benchmark datasets.
Fig 4.
An average predictive accuracy graph using the 10-fold cross-validation technique for threshold value identification.
Fig 5.
An average predictive accuracy graph using training datasets for threshold value identification.
Table 2.
Selected nontext datasets’ characteristics.
Table 3.
Selected text datasets’ characteristics.
Fig 6.
Predictive accuracies of classifiers against benchmark datasets with varying percentages of retained features.
Table 4.
Selected classifier parameters.
Fig 7.
Comparisons of F-measure with existing FS measures.
Table 5.
Comparisons of average classifier precision with existing FS measures.
Table 6.
Comparisons of average classifier recall with existing FS measures.
Table 7.
Comparisons of predictive accuracy (in %age) of the uEFS with existing FS measures.
Table 8.
Paired-samples t-test results.
Table 9.
Comparisons of time measure (in seconds) with existing FS measures.
Table 10.
Comparisons of predictive accuracy (in %age) with existing FS methods.
Table 11.
Comparisons of state-of-the-art ensemble methodologies with the proposed uEFS methodology.
Table 12.
Comparisons of predictive accuracy and F-measure with the Borda method [15].
Table 13.
Comparisons of predictive accuracy and F-measure with the EMFFS method [18].
Fig 8.
Comparisons of F-measure with existing FS measures [29, 37, 39, 48].
Fig 9.
Comparisons of predictive accuracy with existing FS measures [29, 37, 39, 48].
Table 14.
Comparisons of average classifier precision with existing FS methods [29, 37, 39, 48].
Table 15.
Comparisons of average classifier recall with existing FS methods [29, 37, 39, 48].