Replace numpy.NaN with numpy.nan about jpmml-evaluator-python HOT 3 CLOSED

With the NumPy 2.0 release, np.NaN was removed and np.nan should be used instead.

The JPMMl-Evaluator-Python package depends on a higher-level Pandas package (eg. from pandas import DataFrame), not the low(est)-level Numpy package. And as much as I have tested, Pandas 2.2.2 (the latest in PyPI currently) does not work with Numpy 2.0.X.

Of course, this does not mean that one couldn't standardize the representation of Numpy constants in the meantime.

from jpmml-evaluator-python.

And as much as I have tested, Pandas 2.2.2 (the latest in PyPI currently) does not work with Numpy 2.0.X.

This is inconsistent with regards to my work (I am using the latest version of pandas just fine with numpy 2.0 in the codebase for which I have tested), along with what is stated in the pandas documentation here:
https://pandas.pydata.org/docs/whatsnew/v2.2.2.html#pandas-2-2-2-is-now-compatible-with-numpy-2-0

It looks like pandas v2.2.2 does have support for numpy 2.0, and as such, they also do not pin numpy requirements in their project.

Also, to your point, given that those are just constant representations, it probably is worth updating anyways to the newest convention adopted.

from jpmml-evaluator-python.

@vruusmann Thank you for the fix and update!

from jpmml-evaluator-python.