Deploy issue in notebook py_double_ml_pension

[MalteKurz]

Seems to be caused by the new scikit-learn in version 1.0.0, see https://github.com/DoubleML/doubleml-docs/runs/3742278774?check_suite_focus=true.

[MalteKurz]

This is really a bit tricky. Seems to be related to this new check https://github.com/scikit-learn/scikit-learn/blob/6d5774f6895aa84a8ec74762bcb29fc7e5173d41/sklearn/linear_model/_base.py#L728-L782 on Gram matrices introduced here scikit-learn/scikit-learn#19004. As far as I can see np.dot() does not always deliver "numerically" the same result if either called with two 2d-arrays (specific element extracted) or with two 1d-arrays corresponding to that specific element. Unfortunately, I am not (yet) able to extract a minimal working example in order to file a issue to scikit-learn.

[AlejandroTL]

Which is the current state of this issue? I have just suffered it using versions:
scikit-learn 1.2.2
DoubleML 0.6.1

Thanks a lot!

[SvenKlaassen]

The notebook issue should be resolved in the current version. The general sklearn issue with float32 precision might persist, see scikit-learn/scikit-learn#21997