Numpy >= 2.0 not supported

[RobinsonBeaucour]

It seems that numpy >= 2.0 is not supported.

It may be difficult to adapt the code to numpy 2.0. Could it be possible to add numpy<2.0 in the requirements ?

[Niels-c-Scholten]

I am also running into a problem where I get incompatability issues when installing sklearn_extra in a fresh environment.

To give a reproducible method for this incompatability

1. Install a fresh virtual environment with python 3.10
2. pip install scikit-learn-extra
3. run python file import sklearn_extra

This results in the following error: File "sklearn_extra\utils\_cyfht.pyx", line 1, in init sklearn_extra.utils._cyfht ValueError: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject

[marasendra]

Hi @Niels-c-Scholten ,

I reproduced the error with the 3 steps indicated. I was able to bypass this error by downgrading numpy. In your venv you can remove numpy and then execute depending on your dependencies management tool:

• If you are using uv: uv add [email protected]
• If you are using poetry: poetry add [email protected]

While this isn't a long-term fix for the package, it should unblock you and allow you to use scikit-learn-extra immediately. I hope this helps!