Table 5. Average ranks of the classifier calibration errors over all datasets.

Lower values are an indication of possible better performance.

lpICS	lpICS-pre	enICS	enICS-pre	DT	Naive Bayes	SVM-lin	SVM-rbf
4.83	4.83	4.00	4.00	4.17	8.00	2.83	3.83