Closed ksangeek closed 4 years ago
IIRC, we originally used sparse.COO because it works better within dask arrays. If using a scipy.sparse matrix doesn't work, I'd recommend just skipping the test.
Hi @TomAugspurger Thanks for your input. I can make the change to skip this test.
Can you please help me with getting past the CI issue Too long with no output (exceeded 10m0s)
seen in this PR - https://github.com/dask/dask-xgboost/pull/50?
I was able to successfully get the tests run on my local machine!
Thanks, I'm not sure why that would be right now.
This is closed via https://github.com/dask/dask-xgboost/pull/50. Thanks @TomAugspurger!
Describe the bug pytest for
test_sparse()
fails with -Steps/Code to reproduce bug This can be easily reproduced with xgboost 0.82 and 0.90.
Additional context
It looks like XGBoost does not support DMatrix to be created from
sparse.COO
. Looking at the documentation it looks likexgboost.DMatrix(data, ...)
only supports -data (string/numpy.array/scipy.sparse/pd.DataFrame/dt.Frame)
– Data source of DMatrix. When data is string type, it represents the path libsvm format txt file, or binary file that xgboost can read from. ref - https://xgboost.readthedocs.io/en/release_0.90/python/python_api.htmlI see that making this change to use
scipy.sparse.csr_matrix
insteadsparse.COO
helps me get past this issue -