Replies: 5 comments 1 reply
-
I'm not using conda create --name myenv python=3.9 Then you can use Looking at PyPI, it seems that you will need Python 3.8: https://pypi.org/project/scikit-learn/0.23.2/#files Installing from source with Python 3.11 could be challenging because you will most probably need to install from source all the different dependencies as well. |
Beta Was this translation helpful? Give feedback.
-
Well according to PyPi you can use 0.23.2 in Python>=3.6. And I had it working in Python3.9. |
Beta Was this translation helpful? Give feedback.
-
Was the problem ever fixed ? Here's a rough example of how pip install |
Beta Was this translation helpful? Give feedback.
-
Hi @LNkholise Thanks for your interest. I get the following compilation error:
But I can not really decipher what it means, neither how to overcome it. Thanks! |
Beta Was this translation helpful? Give feedback.
-
Okay, let's try this... I can decipher some of the errors so let's start here "fatal error: Python.h: No such file or directory" is due to missing Python development headers, you can resolve that by just installing the python development package |
Beta Was this translation helpful? Give feedback.
-
Hi all,
I have an issue I'm struggling to overcome...
My context: I have an old (from ~2019) machine learning model saved in .pkl and .pb files.
Up until recently I was working with Python 3.9, and Scikit-Learn version 0.23 and the model was working as expected. My venv was fixed to that scikit-learn version because later scikit-learn versions removed sklearn.decomposition.pca and sklearn.mixture.gaussian_mixture (among others) which are used by my model. Further, I can not re-train my model because the training data is long gone...
My issue: In my venv I want know to update Python 3.9 to Python 3.11... but there I can not install scikit-learn==0.23. The output of the pip installation is attached.
error.txt
Could anyone help me on how to install an old scikit-learn version in a new venv with Python 3.11?
PD. I do know the technical correct solution would be to update the machine learning model... but as I said, unfortunately that's is just not possible.
Thank you very much,
Óscar
Beta Was this translation helpful? Give feedback.
All reactions