imblearn.oversampling SMOTENC ValueError

This is my first time using SMOTENC to upsampling my categorical data. However, I’ve been getting error. Can you please advice what should I pass for categorical_features in SMOTENC?

from imblearn.over_sampling import SMOTENC

x=df.drop("A",1)
y=df["A"]

smote_nc = SMOTENC(categorical_features=['A','B','C','D','E','F','G','H'], random_state=0)
X_resampled, y_resampled = smote_nc.fit_resample(x, y)

ERROR:

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-26-f6c9d8a17967> in <module>

---> 14 X_resampled, y_resampled = smote_nc.fit_resample(x, y)
     15 #sm = SMOTE(random_state=100)
     16 #ros = RandomOverSampler()

~/.local/lib/python3.5/site-packages/imblearn/base.py in fit_resample(self, X, y)
     82             self.sampling_strategy, y, self._sampling_type)
     83 
---> 84         output = self._fit_resample(X, y)
     85 
     86         if binarize_y:

~/.local/lib/python3.5/site-packages/imblearn/over_sampling/_smote.py in _fit_resample(self, X, y)
    986     def _fit_resample(self, X, y):
    987         self.n_features_ = X.shape[1]
--> 988         self._validate_estimator()
    989 
    990         # compute the median of the standard deviation of the minority class

~/.local/lib/python3.5/site-packages/imblearn/over_sampling/_smote.py in _validate_estimator(self)
    979                 raise ValueError(
    980                     'Some of the categorical indices are out of range. Indices'
--> 981                     ' should be between 0 and {}'.format(self.n_features_))
    982             self.categorical_features_ = categorical_features
    983         self.continuous_features_ = np.setdiff1d(np.arange(self.n_features_),

ValueError: Some of the categorical indices are out of range. Indices should be between 0 and 7

Answer

As per documentation:

categorical_features : ndarray, shape (n_cat_features,) or (n_features,)
Specified which features are categorical. Can either be:

- array of indices specifying the categorical features;
- mask array of shape (n_features, ) and ``bool`` dtype for which
  ``True`` indicates the categorical features.

So, just replace the line

smote_nc = SMOTENC(categorical_features=['A','B','C','D','E','F','G','H'], random_state=0)

with the line

smote_nc = SMOTENC(categorical_features=[df.dtypes==object], random_state=0)

Advertisement

Answer