'MinMaxScaler' object has no attribute 'clip'
alephpt opened this issue · 0 comments
alephpt commented
I have noticed a bug/deprecation between versions of scikit when trying to run the matcher. This does not appear to be a problem with hmni, but rather version control issues between python, hmni and scikit specifically..
When using Python 11, with various versions of scikit-learn installed, the self.baseModel.predict_proba(x.reshape(1, -1))[0, 1]
line in matcher.py triggers the sklearn pipeline.py which has a dependency to trigger an if self.clip:
check in sklearn\preprocessing_data.py via the predict_proba
File "C:\Users\alephpt\Documents\repos\personal\python\random\string_comp.py", line 6, in hmni_compare
return matcher.similarity(s1, s2)
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\alephpt\AppData\Local\Programs\Python\Python311\Lib\site-packages\hmni\matcher.py", line 266, in similarity
sim = self.meta_inf(pair, features)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\alephpt\AppData\Local\Programs\Python\Python311\Lib\site-packages\hmni\matcher.py", line 418, in meta_inf
meta_features[0] = self.base_model_inf(base_features)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\alephpt\AppData\Local\Programs\Python\Python311\Lib\site-packages\hmni\matcher.py", line 413, in base_model_inf
y_pred = self.baseModel.predict_proba(x.reshape(1, -1))[0, 1]
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\alephpt\AppData\Local\Programs\Python\Python311\Lib\site-packages\sklearn\pipeline.py", line 546, in predict_proba
Xt = transform.transform(Xt)
^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\alephpt\AppData\Local\Programs\Python\Python311\Lib\site-packages\sklearn\utils\_set_output.py", line 140, in wrapped
data_to_wrap = f(self, X, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\alephpt\AppData\Local\Programs\Python\Python311\Lib\site-packages\sklearn\preprocessing\_data.py", line 518, in transform
if self.clip:
^^^^^^^^^
I believe one solution would be to set self.clip to none for transform instance, but I'm not sure how to do this from my code without trying to change the matcher.py instance in hmni 🤔