This pull request modifies the OpenmlSource and OpenmlSourceTests which previously accommodated CSVs only. To test non-sparse ARFFs use test_default_classification from OpenmlArffSource_Tests in test_simulations.py.
To test sparse data use test_sparse_classification from OpenmlArffSource_Tests in test_simulations.py. Note that the sparse data implementation is inefficient and will require optimization.
Classes created in data/filters.py to handle ARFF data:
This pull request modifies the
OpenmlSourceandOpenmlSourceTestswhich previously accommodated CSVs only. To test non-sparse ARFFs usetest_default_classificationfromOpenmlArffSource_Testsintest_simulations.py.To test sparse data use
test_sparse_classificationfromOpenmlArffSource_Testsintest_simulations.py. Note that the sparse data implementation is inefficient and will require optimization.Classes created in
data/filters.pyto handle ARFF data:sparse_fillermethod handles sparse data)