CODEWITHSUNDEEP
pythonnumpymachine-learningscikit-learnone-hot-encoding
Daniil Orekhov
Daniil Orekhov

Reputation: 438

OneHotEncoder from sklearn gives a ValueError when passing categories

I have an array of class names:

classes = np.array(['A', 'B'])

And I have an array of data (but this data only contains instances of one class):

vals = np.array(['A', 'A', 'A'])
vals = vals.reshape(len(vals), 1)

I want to end up with one-hot encoding for the vals array, such that it accounts for the fact that there might be some other classes. I am trying to use sklearn.preprocessing.OneHotEncoder:

ohe = OneHotEncoder(sparse=False, categories=classes)
ohe.fit_transform(vals)

But when I run this, I get the following error:

Traceback (most recent call last):
  File "/usr/local/anaconda3/envs/my_project/lib/python3.6/site-packages/IPython/core/interactiveshell.py", line 3331, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-10-08d325b5e8a7>", line 1, in <module>
    ohe.fit_transform(vals)
  File "/usr/local/anaconda3/envs/my_project/lib/python3.6/site-packages/sklearn/preprocessing/_encoders.py", line 372, in fit_transform
    return super().fit_transform(X, y)
  File "/usr/local/anaconda3/envs/my_project/lib/python3.6/site-packages/sklearn/base.py", line 571, in fit_transform
    return self.fit(X, **fit_params).transform(X)
  File "/usr/local/anaconda3/envs/my_project/lib/python3.6/site-packages/sklearn/preprocessing/_encoders.py", line 347, in fit
    self._fit(X, handle_unknown=self.handle_unknown)
  File "/usr/local/anaconda3/envs/my_project/lib/python3.6/site-packages/sklearn/preprocessing/_encoders.py", line 76, in _fit
    if self.categories != 'auto':
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

Upvotes: 3

Views: 274

Answers (1)

FBruzzesi
FBruzzesi

Reputation: 6475

You can fit the encoder with classes and then trasform vals:

import numpy as np
from sklearn.preprocessing import OneHotEncoder

classes = np.array(['A', 'B'])
vals = np.array(['A', 'A', 'A'])
vals = vals.reshape(-1, 1)

ohe = OneHotEncoder(sparse=False)
ohe.fit(classes.reshape(-1, 1))

ohe.transform(vals)
array([[1., 0.],
       [1., 0.],
       [1., 0.]])

Upvotes: 4

Related Questions