ARFF datasets

WEKA datasets
Other collection

The ELF reader for ARFF files supports only categorical features, where all entries are defined in the attribute section. For example when the value '?' occur in the data section and it is not defined for this attribute, the data-readin would fail.

The file settings.txt contains the dataset name of train and test set and the name of the target column. Please note that the test data must also contain target values. For example:
train=UCI/diabetes.arff
test=UCI/diabetesTest.arff
trainTargetColumn='class'

The ARFF reader works for the following datasets from UCI WEKA datasets (first jar file from this page).
We have a preconfigured directory with arff files here.