Shown below are the cross-validation balanced error rates of UCI datasets of a single 90:10 train validation split (trainlabels.0 in our datasets repository). Each method is run on the original dataset.

Datasets	Nearest means	Naive bayes	Least squares adapative eta (term=.001)	Hinge loss adaptive eta (term=.001)
Breast cancer	15.9	8.7	20.5	13.3
Climate	20	30	50	10
Qsar	33.3	17.4	23	15.4
Ionosphere	36.6	34.8	25.3	17.6
Micromass	29.5	17.5	50	10.3
Hill valley	46.8	51.6	50	27.4
Average	30.4	51.6	36.5	15.7