Feature names from OneHotEncoder

Question

I am using OneHotEncoder to encode few categorical variables (eg - Sex and AgeGroup). The resulting feature names from the encoder are like - 'x0_female', 'x0_male', 'x1_0.0', 'x1_15.0' etc.

>>> train_X = pd.DataFrame({'Sex':['male', 'female']*3, 'AgeGroup':[0,15,30,45,60,75]})

>>> from sklearn.preprocessing import OneHotEncoder
>>> encoder = OneHotEncoder()
>>> train_X_encoded = encoder.fit_transform(train_X[['Sex', 'AgeGroup']])

>>> encoder.get_feature_names()
>>> array(['x0_female', 'x0_male', 'x1_0.0', 'x1_15.0', 'x1_30.0', 'x1_45.0',
       'x1_60.0', 'x1_75.0'], dtype=object)

Is there a way to tell OneHotEncoder to create the feature names in such a way that the column name is added at the beginning, something like - Sex_female, AgeGroup_15.0 etc, similar to what Pandas get_dummies() does.

Answer 1

A list with the original column names can be passed to get_feature_names.

>>> encoder.get_feature_names(['Sex', 'AgeGroup'])

array(['Sex_female', 'Sex_male', 'AgeGroup_0', 'AgeGroup_15',
       'AgeGroup_30', 'AgeGroup_45', 'AgeGroup_60', 'AgeGroup_75'],
      dtype=object)

• DEPRECATED: get_feature_names is deprecated in 1.0 and will be removed in 1.2. Please use get_feature_names_out instead.
  • As per sklearn.preprocessing.OneHotEncoder.

>>> encoder.get_feature_names_out(['Sex', 'AgeGroup'])

array(['Sex_female', 'Sex_male', 'AgeGroup_0', 'AgeGroup_15',
       'AgeGroup_30', 'AgeGroup_45', 'AgeGroup_60', 'AgeGroup_75'],
      dtype=object)

Answer 2

• type(train_X_encoded) → scipy.sparse.csr.csr_matrix
  • Use pandas.DataFrame.sparse.from_spmatrix to load a sparse matrix, otherwise convert to a dense matrix and load with pandas.DataFrame.

# pandas.DataFrame.sparse.from_spmatrix will load this sparse matrix
>>> print(train_X_encoded)

  (0, 1)    1.0
  (0, 2)    1.0
  (1, 0)    1.0
  (1, 3)    1.0
  (2, 1)    1.0
  (2, 4)    1.0
  (3, 0)    1.0
  (3, 5)    1.0
  (4, 1)    1.0
  (4, 6)    1.0
  (5, 0)    1.0
  (5, 7)    1.0

# pandas.DataFrame will load this dense matrix
>>> print(train_X_encoded.todense())

[[0. 1. 1. 0. 0. 0. 0. 0.]
 [1. 0. 0. 1. 0. 0. 0. 0.]
 [0. 1. 0. 0. 1. 0. 0. 0.]
 [1. 0. 0. 0. 0. 1. 0. 0.]
 [0. 1. 0. 0. 0. 0. 1. 0.]
 [1. 0. 0. 0. 0. 0. 0. 1.]]

import pandas as pd

column_name = encoder.get_feature_names_out(['Sex', 'AgeGroup'])
one_hot_encoded_frame = pd.DataFrame.sparse.from_spmatrix(train_X_encoded, columns=column_name)

# display(one_hot_encoded_frame)
   Sex_female  Sex_male  AgeGroup_0  AgeGroup_15  AgeGroup_30  AgeGroup_45  AgeGroup_60  AgeGroup_75
0         0.0       1.0         1.0          0.0          0.0          0.0          0.0          0.0
1         1.0       0.0         0.0          1.0          0.0          0.0          0.0          0.0
2         0.0       1.0         0.0          0.0          1.0          0.0          0.0          0.0
3         1.0       0.0         0.0          0.0          0.0          1.0          0.0          0.0
4         0.0       1.0         0.0          0.0          0.0          0.0          1.0          0.0
5         1.0       0.0         0.0          0.0          0.0          0.0          0.0          1.0

• From scikit-learn v1.0 use get_feature_names_out instead of get_feature_names

Answer 3

Thanks for a nice solution. @Nursnaaz The sparse matrix needs to convert into a dense matrix.

column_name = encoder.get_feature_names(['Sex', 'AgeGroup'])
one_hot_encoded_frame =  pd.DataFrame(train_X_encoded.todense(), columns= column_name)

• From scikit-learn v1.0 use get_feature_names_out instead of get_feature_names