Skip to main content

Table 3 Results of Ten-cross-fold validation on RSODS by WEKA

From: Harnessing ontology and machine learning for RSO classification

EI M&D
R-OBC J48 BayesNet
O[1] O[2] R S A O R S A O R S A
Accuracy 0.872 0.85 0.893 0.892 0.884 0.854 0.837 0.707 0.848 0.899 0.862 0.862 0.878
(W)FP rate 0.03 0.041 0.033 0.032 0.031 0.124 0.157 0.437 0.119 0.027 0.14 0.137 0.085
(W)precision 0.891 0.87 0.892 0.892 0.891 0.809 0.79 0.609 0.807 0.904 0.862 0.864 0.876
(W)recall 0.891 0.87 0.894 0.893 0.892 0.854 0.837 0.707 0.848 0.899 0.862 0.862 0.878
T (s) 0.87 4.1 0.36 0.29 0.46 3.74 3.58 3.64 3.95 0.27 0.11 0.11 0.13
EI M&D
JRip SMO RandomForest (50-trees)
O R S A O R S A O R S A
Accuracy 0.92 0.864 0.817 0.838 0.903 0.888 0.395 0.903 0.9 0.889 0.881 0.893
(W)FP rate 0.035 0.084 0.228 0.191 0.04 0.044 0.251 0.041 0.082 0.103 0.136 0.082
(W)precision 0.918 0.847 0.816 0.843 0.902 0.886 0.527 0.902 0.897 0.888 0.879 0.892
(W)recall 0.92 0.864 0.817 0.838 0.903 0.888 0.395 0.903 0.9 0.889 0.881 0.893
T (s) 3.7 3.23 3.31 3.64 3.67 3.47 3.65 3.70 313.52 271.91 277.67 297.64
EI M&D
MultilayerPerceptron (4-hidden-layer) SimpleLogistic
O R S A O R S A
Accuracy 0.847 0.46 0.155 0.740 0.9 0.897 0.746 0.742
(W)FP rate 0.044 0.084 0.085 0.223 0.039 0.048 0.267 0.373
(W)precision 0.813 0.691 0.667 0.668 0.899 0.901 0.683 0.635
(W)recall 0.847 0.46 0.155 0.740 0.9 0.897 0.746 0.742
T (s) 294.83 342.32 308.75 315.10 3532.5 3440.7 3479 3643.57
  1. Italic values indicate moderate negative significance
  2. Underlined values indicate significant negative impact
  3. Normal values indicate trivial significance
  4. M&D: methods and data; (W)EI: (weighted average) evaluation index; O: original test data; R: unknown rcs in test data; S: unknown size in test data; A: unknown amr in test data; T: learning time for building model