Skip to main content

Advertisement

Table 2 Comparison of classification algorithms according to a number of correctly classified elements, false positive, false negative and the classifiers accuracies

From: Using Pareto points for model identification in predictive toxicology

Method Correct class Falsepositive False negative Accuracy
SMO 899 122 (10.8%) 106 (9.4%) 0.80
Part 904 123 (10.9%) 101 (8.9%) 0.80
NaiveBayes 845 191 (19%) 90 (7.9%) 0.75
J48 905 123 (10.9%) 100 (8.9%) 0.80
IBK(1) 905 121 (10.7%) 102 (9%) 0.80
IBK(3) 901 133 (11.7%) 94 (8.3%) 0.79
IBK(5) 889 149 (13.2%) 93(8.2%) 0.78
BayesNet 756 264 (23%) 108 (9.5%) 0.67
DMS 901 115 (10.1%) 112 (9.9%) 0.79
3-CPMI 902 136 (12%) 90 (7.9%) 0.79
5-CPMI 897 137 (12%) 94 (8.3%) 0.79
10-CPMI 863 168 (14.8%) 97 (8.5%) 0.76
3-APMI 918 99 (8.7%) 111(9.8%) 0.81
5-APMI 891 115 (10%) 122 (10.8%) 0.78
  1. The polar narcosis model label was defined as the positive class.