MultiTaskLassoCV got an unexpected keyword argument 'normalize'

[anhtvan]

Hi, I'm getting an issue when trying to run this code. As far as I know the data is correclty formatted. The type error relates to some LassCV backend - could someone help me please?

I am running it on both python 3.7.9 and 3.11.2, so hopefully it's not a python issue.

[sak-18]

Same issue. Found any root cause anhtvan?

[sak-18]

https://scikit-learn.org/1.1/modules/generated/sklearn.linear_model.MultiTaskLassoCV.html?highlight=multitasklassocv Looking at this doc seems like normalize argument was deprecated and later removed altogether from scikit-learn 1.2. Downgrading sci-kit-learn should do, I hope for now.

[tom-flamelit]

Saketh Vishnubhatla (@sak-18) which version of scikit-learn do you recommend?

[samuelgarca]

Has anyone found a solution, I tried to revert Scikit-learn to version 1.0, but I was unsuccessful.

[VanAltrades]

Samuel Garcia (@samuelgarca)
I just ran into this unexpected 'normalize' argument as well.

I cannot confirm the root cause, but I was able to solve this error by modifying the _MTLassoCV_MatchSpace function.
I accessed the function via this path:

All I did was comment out the lines containing normalize=normalize and replace them with the same line sans normalize=normalize.

Since this modification, I have been able to run SparseSC.fit_fast without errors.

I'm not knowledgeable enough about versioning and compatibility to explain the root cause, but hopefully, this tip will help you get the package operational.

Jack

[MJeremy2017]

I think the sklearn dependency is obsolete. In the newer version this parameter is already removed

[franc703]

I have done the same as VanAltrades (@VanAltrades) and run the code without any problems. Make sure you normalize the data before.