OneHotEncoder：__init__() 得到了一个意外的关键字参数“categorical_features”

Question

我在使用 onehotencoder 时遇到此错误，其中缺少 thecategorical_features 属性，我正在使用 google colab。

from sklearn.preprocessing import LabelEncoder, OneHotEncoder
le = LabelEncoder()
X = star.iloc[:,:6].values
y = star.iloc[:,-1].values
X[:,5] = le.fit_transform(X[:,5])
y[:] = le.fit_transform(y[:])

ohe = OneHotEncoder(categorical_features= [5])
X = ohe.fit_transform(X).toarray()


TypeError                                 Traceback (most recent call last)
<ipython-input-47-93f73a1a04ad> in <module>()
----> 1 ohe = OneHotEncoder(categorical_features= [5])
      2 X = ohe.fit_transform(X).toarray()

TypeError: __init__() got an unexpected keyword argument 'categorical_features'

Answer 1

在OneHotEncoder 的文档中没有'categorical_features'

OneHotEncoder 的旧文档 (0.20)显示'categorical_features'将在 0.22 中删除（sklearn最新版本的编号为 0.22.1），您必须使用ColumnTransformer。

但我不知道如何使用它。但也许用户指南中的示例可以帮助使用它。

编辑：

sklearn 0.20

pip install -U scikit-learn==0.20

代码

import sklearn
print(sklearn.__version__)

from sklearn.preprocessing import LabelEncoder, OneHotEncoder
import numpy as np

print('--- data ---')
X = [
    ['Male',   1],
    ['Female', 3],
    ['Female', 2]
]
X = np.array(X)
print(X)

print('--- LabelEncoder ---')
le = LabelEncoder()
X[:,0] = le.fit_transform(X[:,0])
print(X)

print('--- OneHotEncoder ---')
ohe = OneHotEncoder(categorical_features=[0])
X = ohe.fit_transform(X).toarray()
print(X)

结果：

0.20.0
--- data ---
[['Male' '1']
 ['Female' '3']
 ['Female' '2']]
--- LabelEncoder ---
[['1' '1']
 ['0' '3']
 ['0' '2']]
--- OneHotEncoder ---
[[0. 1. 1.]
 [1. 0. 3.]
 [1. 0. 2.]]

sklearn 0.22

pip install -U scikit-learn==0.22

代码：

import sklearn
print(sklearn.__version__)

from sklearn.preprocessing import LabelEncoder, OneHotEncoder
from sklearn.compose import ColumnTransformer
import numpy as np

print('--- data ---')
X = [
    ['Male',   1],
    ['Female', 3],
    ['Female', 2]
]
X = np.array(X)
print(X)

print('--- LabelEncoder ---')
le = LabelEncoder()
X[:,0] = le.fit_transform(X[:,0])
print(X)

print('--- OneHotEncoder ---')
ct = ColumnTransformer([('my_ohe', OneHotEncoder(), [0])], remainder='passthrough')
X = ct.fit_transform(X) #.toarray()
print(X)

结果：

0.22.2.post1
--- data ---
[['Male' '1']
 ['Female' '3']
 ['Female' '2']]
--- LabelEncoder ---
[['1' '1']
 ['0' '3']
 ['0' '2']]
--- OneHotEncoder ---
[['0.0' '1.0' '1']
 ['1.0' '0.0' '3']
 ['1.0' '0.0' '2']]

在 0.22 中，即使没有LabelEncoder但 0.20 的需求也能工作LabelEncoder

import sklearn
print(sklearn.__version__)

from sklearn.preprocessing import OneHotEncoder
from sklearn.compose import ColumnTransformer
import numpy as np

print('--- data ---')
X = [
    ['Male',   1],
    ['Female', 3],
    ['Female', 2]
]
X = np.array(X)
print(X)

print('--- OneHotEncoder ---')
ct = ColumnTransformer([('my_ohe', OneHotEncoder(), [0])], remainder='passthrough')
X = ct.fit_transform(X) #.toarray()
print(X)