search

functime-org / functime

Time-series machine learning at scale. Built with Polars for embarrassingly parallel feature extraction and forecasts on panel data.
https://docs.functime.ai
Apache License 2.0
971 stars 52 forks source link

test failures #187

Closed baggiponte closed 4 weeks ago

baggiponte commented 3 months ago

if I run pytest --benchmark-skip I get these errors.

FAILED tests/test_preprocessing.py::test_yeojohnson[n_periods(50)-n_entities(10)] - polars.exceptions.ComputeError: TypeError: No loop matching the specified signature and casting was found for ufunc isfinite
FAILED tests/test_tsfresh.py::test_mean_abs_change[S9-res9-k9] - polars.exceptions.InvalidOperationError: `abs` operation not supported for dtype `null`
FAILED tests/test_tsfresh.py::test_absolute_maximum[S0-res0] - AssertionError: columns ['a'] in left DataFrame, but not in right
FAILED tests/test_tsfresh.py::test_absolute_maximum[S1-res1] - AssertionError: columns ['a'] in left DataFrame, but not in right
FAILED tests/test_tsfresh.py::test_absolute_maximum[S2-res2] - AssertionError: columns ['a'] in left DataFrame, but not in right
FAILED tests/test_tsfresh.py::test_binned_entropy[S0-res0-10] - AssertionError: columns ['count'] in left DataFrame, but not in right
FAILED tests/test_tsfresh.py::test_binned_entropy[S1-res1-10] - AssertionError: columns ['count'] in left DataFrame, but not in right
FAILED tests/test_tsfresh.py::test_binned_entropy[S2-res2-100] - AssertionError: columns ['count'] in left DataFrame, but not in right
FAILED tests/test_tsfresh.py::test_ratio_beyond_r_sigma_nan_case[S0-res0-1] - polars.exceptions.InvalidOperationError: `std` operation not supported for dtype `null`
FAILED tests/test_tsfresh.py::test_root_mean_square_nan_case[S0-res0] - polars.exceptions.InvalidOperationError: `sum` operation not supported for dtype `null`
FAILED tests/test_tsfresh.py::test_mean_n_absolute_max[S0-1-res0] - polars.exceptions.InvalidOperationError: `abs` operation not supported for dtype `null`
baggiponte commented 3 months ago

@marcogorelli sorry to bother, do you have any clue what this is? polars.exceptions.ComputeError: TypeError: No loop matching the specified signature and casting was found for ufunc isfinite

baggiponte commented 1 month ago

@abstractqqq sorry to ping. Just ran the tests again after your amazing PR and there's only one that fails:

FAILED tests/test_preprocessing.py::test_diff[n_periods(50)-n_entities(10)-1] - AssertionError: DataFrames are different (value mismatch for column 'open')

Can you have a look at that, please? Then we can close this issue. Thank you very much.

MarcoGorelli commented 1 month ago

@MarcoGorelli sorry to bother, do you have any clue what this is? polars.exceptions.ComputeError: TypeError: No loop matching the specified signature and casting was found for ufunc isfinite

sorry, forgot about this one - are you applying a numpy ufunc directly on an expression somewhere?

baggiponte commented 1 month ago

sorry, forgot about this one - are you applying a numpy ufunc directly on an expression somewhere?

Oh don't worry. I think we are. Should double check that. However TQ solved 99% of those recently 😎

baggiponte commented 4 weeks ago

Tests now pass! Closing the issue.