AttributeError: module 'sklearn.metrics' has no attribute 'jaccard_similarity_score'

[petraknovak]

I am using scikit-learn version 0.23.1 and I get the following error:
AttributeError: module 'sklearn.metrics' has no attribute 'jaccard_similarity_score'
when calling the function ConfusionMatrix.

[guitarzero]

Same here. Thanks

[blaze515]

I'm also getting this error...

It looks like jaccard_similarity_score was deprecated since sckiti-learn 0.21 in favor of newer implementation, jaccard_score, and was removed in 0.23:

https://scikit-learn.org/0.22/modules/generated/sklearn.metrics.jaccard_similarity_score.html?highlight=jaccard#sklearn.metrics.jaccard_similarity_score

scikit-learn/scikit-learn#13151

[aparente-nurix]

Same here, there's also a stack overflow for a different error that suggests this is an sklearn version issue.

https://stackoverflow.com/questions/61819628/is-pandas-ml-broken

[benlevitas]

Looks like they changed the sklearn function to 'jaccard_score'
https://scikit-learn.org/stable/modules/generated/sklearn.metrics.jaccard_score.html

[hawk-praxs]

Agree with @benlevitas if you want the library to function correctly, make the following change:
In ~/anaconda3/Lib/site-packages/pandas_ml/skaccessors/metrics.py
Line 249: change jaccard_similarity_score to jaccard_score

The _classification_methods list has the list of accuracy metrics. Changing it leads to the proper functioning of the library. I hope they update this in the file as well.

[ipesanz]

In addition to what @hawk-praxs mentioned, then another dependency change will appear related to Imputer, and then another.. an so [. . .]

AttributeError: module 'sklearn.preprocessing' has no attribute 'Imputer'

[hawk-praxs]

In addition to what @hawk-praxs mentioned, then another dependency change will appear related to Imputer, and then another.. an so [. . .]

AttributeError: module 'sklearn.preprocessing' has no attribute 'Imputer'

Fixing the Imputer class in the library makes it work.

No further errors after this.

[ipesanz]

I changed on file:
[..]/lib/python3.8/site-packages/pandas_ml/skaccessors/preprocessing.py the only coincidence to Imputer.

pp.imputer >> pp.SimpleImputer [Bear in Mind that SimpleImputer is no longer in preprocessing, is in "Imputer" class]

New in version 0.20: SimpleImputer replaces the previous sklearn.preprocessing.Imputer estimator which is now removed.

From the SkLearn docs

Error after:

[averner]

SimpleImputer is now in sklearn.impute (https://scikit-learn.org/stable/modules/generated/sklearn.impute.SimpleImputer.html). My solution was to change preprocessing.py as followins:
...
if _SKLEARN_INSTALLED:
import sklearn.preprocessing as pp
from sklearn.impute import SimpleImputer
_keep_col_classes = [pp.Binarizer,
pp.FunctionTransformer,
SimpleImputer,
...

I still get "AttributeError: type object 'NDFrame' has no attribute 'groupby'" after these changes though.

[robertson809]

How has this not been fixed in the pandas ml library yet?

[averner]

Sorry, I've no idea. I'm not using it anymore. пн, 19 дек. 2022 г. в 07:54, Michael Robertson ***@***.***>:

…

How has this not been fixed in the pandas ml library yet? — Reply to this email directly, view it on GitHub <#131 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AQU33YI4R6IBOZ6EIEQNAZTWN7Z7XANCNFSM4ORT4PUQ> . You are receiving this because you commented.Message ID: ***@***.***>

[olchemist]

I got a new one concerning seaborn.factorplot, so I changed ~/opt/anaconda3/lib/python3.9/site-packages/pandas_ml/snsaccessors/base.py

replace "factorplot" with "catplot" in line 223