Identification of a minimum number of genes to predict triple-negative breast cancer subgroups from gene expression profiles

Table 1 Comparative overview of seven prediction algorithms according to the 120 upregulated genes

	TP rate	FP rate	Accuracy %	Mean absolute error	Kappa	Precision	Recall	F-measure	MCC	ROC area	PRC area
Naive Bayes	0.863	0.035	86.3291	0.0452	0.8281	0.864	0.863	0.863	0.828	0.981	0.934
Logistic regression	0.595	0.105	59.4937	0.1343	0.4884	0.596	0.595	0.594	0.492	0.862	0.664
Multilayer perceptron	0.894	0.027	89.3671	0.0429	0.8658	0.894	0.894	0.893	0.867	0.987	0.951
Support vector machine	0.889	0.029	88.8608	0.2257	0.8597	0.889	0.889	0.888	0.860	0.963	0.845
k-Nearest neighbours	0.808	0.049	80.7595	0.0677	0.7579	0.811	0.808	0.808	0.759	0.872	0.695
Decision tree	0.646	0.096	64.557	0.1414	0.5488	0.653	0.646	0.646	0.557	0.845	0.603
Random rorest	0.858	0.046	85.8228	0.134	0.8191	0.865	0.858	0.852	0.821	0.985	0.941

TP, true positive; FP, false positive; MCC, Matthews correlation coefficient; ROC, relative operating characteristic; PRC precision–recall curve

ISSN: 1479-7364