`np.NINF` was removed in the NumPy 2.0 release. Use `-np.inf` instead

[tam-nguyenminh]

Hello,

I got this error while trying to use SequentialFeatureSelector. Could you please help take a look in the underlying codes? Thanks a lot

sfs = SFS(LogisticRegression(), k_features=5, #number of features to select forward = True, #rolling forward selection floating=False, # No floating selection scoring='accuracy', verbose=5, cv=5 ).fit(X_train,y_train)

---

AttributeError Traceback (most recent call last) Cell In[214], line 8 1 sfs = SFS(LogisticRegression(), 2 k_features=5, #number of features to select 3 forward = True, #rolling forward selection 4 # floating=False, # No floating selection 5 scoring='accuracy', 6 verbose=5, 7 cv=5 ----> 8 ).fit(X_train,y_train)

File c:\Users\84393\miniconda3\envs\venv_ds\Lib\site-packages\mlxtend\feature_selection\sequential_feature_selector.py:650, in SequentialFeatureSelector.fit(self, X, y, groups, **fit_params) 648 else: 649 self.fitted = True # the completion of sequential selection process. --> 650 self.finalize_fit() 652 return self

File c:\Users\84393\miniconda3\envs\venv_ds\Lib\site-packages\mlxtend\feature_selection\sequential_feature_selector.py:655, in SequentialFeatureSelector.finalize_fit(self) 654 def finalize_fit(self): --> 655 max_score = np.NINF 656 for k in self.subsets_: 657 if ( 658 k >= self.min_k 659 and k <= self.max_k 660 and self.subsets_[k]["avg_score"] > max_score 661 ):

File c:\Users\84393\miniconda3\envs\venv_ds\Lib\site-packages\numpy__init.py:411, in getattr(attr) 408 raise AttributeError(__former_attrs[attr]) 410 if attr in expired_attributes__: --> [411](file:///C:/Users/84393/miniconda3/envs/venv_ds/Lib/site-packages/numpy/init.py:411) raise AttributeError( 412 f"np.{attr} was removed in the NumPy 2.0 release. " [413](file:///C:/Users/84393/miniconda3/envs/venv_ds/Lib/site-packages/numpy/init.py:413) f"{expired_attributes[attr]}" 414 ) [416](file:///C:/Users/84393/miniconda3/envs/venv_ds/Lib/site-packages/numpy/init.py:416) if attr == "chararray": 417 warnings.warn( [418](file:///C:/Users/84393/miniconda3/envs/venv_ds/Lib/site-packages/numpy/init.py:418) "np.chararray is deprecated and will be removed from " 419 "the main namespace in the future. Use an array with a string " [420](file:///C:/Users/84393/miniconda3/envs/venv_ds/Lib/site-packages/numpy/init__.py:420) "or bytes dtype instead.", DeprecationWarning, stacklevel=2)

AttributeError: np.NINF was removed in the NumPy 2.0 release. Use -np.inf instead.

[Ontheroad123]

modofied 655 "max_score = np.NINF" to "max_score = np.inf"

[shobhit120801]

modofied 655 "max_score = np.NINF" to "max_score = np.inf"

do we have to change it manually ?

[cwoodside1278]

Hey @shobhit120801 I ran into the exact same error as you. @Ontheroad123 I did update the source code manually to np.inf or even -np.inf and neither work. I keep getting the same error above even though I have updated the source code

[Keeva1]

I don't know anything about codes Brad

On Tue, Oct 15, 2024, 3:04 PM Christie-Rose Woodside < @.***> wrote:

Hey @shobhit120801 https://github.com/shobhit120801 I ran into the exact same error as you. @Ontheroad123 https://github.com/Ontheroad123 I did update the source code manually to np.inf or even -np.inf and neither work. I keep getting the same error above even though I have updated the source code

— Reply to this email directly, view it on GitHub https://github.com/rasbt/mlxtend/issues/1100#issuecomment-2414901197, or unsubscribe https://github.com/notifications/unsubscribe-auth/BFXOEMCEV6JAV6ZLAVGDA3DZ3VYNPAVCNFSM6AAAAABJ3MTS7WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMJUHEYDCMJZG4 . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[d-kleine]

Hey @shobhit120801 I ran into the exact same error as you. @Ontheroad123 I did update the source code manually to np.inf or even -np.inf and neither work. I keep getting the same error above even though I have updated the source code

@cwoodside1278 It worked for me, but you need to replace two lines in the source code:

 def finalize_fit(self):
        max_score = -np.inf # changed
        ...
        k_score = max_score
        if k_score == -np.inf: # changed
        ...

If anyone of you needs help with this, let me know. I can also push a PR on this.

[baimamboukar]

What finally worked for you guys ?

I ran into the exact same issue as @tam-nguyenminh

sfs_forward = SFS(linear_model,
                  k_features='best',
                  forward=True,
                  floating=False,
                  scoring='neg_mean_squared_error',
                  cv=5)

sfs_forward = sfs_forward.fit(predictor_variables, dependent_variable) # ---> 
#🔴This line has an error (AttributeError: `np.NINF` was removed in the NumPy 2.0 release. Use `-np.inf` instead.Cell Execution Error)

I am not sure what to update to make it work. @d-kleine

[d-kleine]

You need to edit the source code, as provided above.

Clone the repo locally, apply the changes - or checkout my PR at gh pr checkout 1107 locally. Then install the repo locally, e.g. pip install ./mlxtend (open a command shell, activate the relevant Python env, nagivate to the working dir where the cloned mlxtend folder is located in, e.g. your Desktop, and then execute the command)

[d-kleine]

A new version of mlxtend has just been released (0.23.2), you can update it via pip install --upgrade mlxtend. Once updated, above error message should not pop up anymore.