Table 5.

Colon cancer prediction rankings

Classifier	MCC	F1 score	Accuracy	TP rate	TN rate
MCC ranking:
Gradient boosting	+0.55	0.81	0.78	0.85	0.69
Decision tree	+0.53	0.82	0.77	0.88	0.58
k-nearest neighbors	+0.48	0.87	0.80	0.92	0.52
Linear SVM	+0.41	0.82	0.76	0.86	0.53
Radial SVM	+0.29	0.75	0.67	0.86	0.40
F1 score ranking:
k-nearest neighbors	+0.48	0.87	0.80	0.92	0.52
Linear SVM	+0.41	0.82	0.76	0.86	0.53
Decision tree	+0.53	0.82	0.77	0.88	0.58
Gradient boosting	+0.55	0.81	0.78	0.85	0.69
Radial SVM	+0.29	0.75	0.67	0.86	0.40
Accuracy ranking:
k-nearest neighbors	+0.48	0.87	0.80	0.92	0.52
Gradient boosting	+0.55	0.81	0.78	0.85	0.69
Decision tree	+0.53	0.82	0.77	0.88	0.58
Linear SVM	+0.41	0.82	0.76	0.86	0.53
Radial SVM	+0.29	0.75	0.67	0.86	0.40

Prediction results on colon cancer gene expression dataset, based on MCC, F₁ score, and accuracy. linear SVM: support vector machines with linear kernel. MCC: worst value –1 and best value +1. F₁ score, accuracy, TP rate, and TN rate: worst value 0 and best value 1. To avoid additional complexity and keep this table simple to read, we prefered to exclude the standard deviation of each result metric. We highlighted in bold the ranking of each rate