lightgbm

Summary

mypy shows some issues in LightGBM's Python package.

mypy \
    --exclude='python-package/compile/|python-package/build' \
    --ignore-missing-imports \
    python-package/

python-package/lightgbm/compat.py:12: error: Name 'Series' already defined (possibly by an import)
python-package

I have a simple regression task (using a LightGBMRegressor) where I want to penalize negative predictions more than positive ones. Is there a way to achieve this with the default regression LightGBM objectives (see https://lightgbm.readthedocs.io/en/latest/Parameters.html)? If not, is it somehow possible to define (many example for default LightGBM model) and pass a custom regression objective?

Describe the bug

Failed to execute Series.drop_duplicates.

In [75]: a = md.DataFrame(np.random.rand(10, 2), columns=['a', 'b'], chunk_size=2)                  

In [76]: a['a'].drop_duplicates().execute()               

Our xgboost models use the binary:logistic' objective function, however the m2cgen converted version of the models return raw scores instead of the transformed scores.

This is fine as long as the user knows this is happening! I didn't, so it took a while to figure out what was going on. I'm wondering if perhaps a useful warning could be raised for users to alert them of this issue? A warning

Details in discussion mljar/mljar-supervised#421

Currently we don't test (or document) that Eland should work with data streams, we should probably test that everything works properly.