OneHotEncoder

参数

def __init__(self, n_values=None, categorical_features=None,
                 categories=None, sparse=True, dtype=np.float64,
                 handle_unknown='error'):

handle_unknown

数据在转化为 one-hot 编码时，如果遇到一个属性值没有在训练集中出现，程序应该怎么办？如果是 error 的话，程序就报错；如果是 ignore 的话，程序忽略可以继续执行。

from sklearn.preprocessing import OneHotEncoder
enc = OneHotEncoder(handle_unknown='ignore')
x_train = [['Male', 1], ['Female', 3], ['Female', 2]]
enc.fit(x_train)
# [array(['Female', 'Male'], dtype=object), array([1, 2, 3], dtype=object)]
print(enc.categories_)
x_test = [['neutral', 1], ['Male', 4]]
# [[0. 0. 1. 0. 0.]
#  [0. 1. 0. 0. 0.]]
x_test = enc.transform(x_test).toarray()
print(x_test)

sparse

如果是 True，则返回 sparse matrix，如果为 False，则返回 an array。

from sklearn.preprocessing import OneHotEncoder
enc = OneHotEncoder(handle_unknown='ignore')
x_train = [['Male', 1], ['Female', 3], ['Female', 2]]
enc.fit(x_train)
x_test = [['neutral', 1], ['Male', 4]]
x_test = enc.transform(x_test)
#   (0, 2)  1.0
#   (1, 1)  1.0
print(x_test)

enc = OneHotEncoder(handle_unknown='ignore',sparse=False)
x_train = [['Male', 1], ['Female', 3], ['Female', 2]]
enc.fit(x_train)
x_test = [['neutral', 1], ['Male', 4]]
x_test = enc.transform(x_test)
# [[0. 0. 1. 0. 0.]
#  [0. 1. 0. 0. 0.]]
print(x_test)

查看每一列属性种类

• categories_
• get_feature_names

# [array(['Female', 'Male'], dtype=object), array([1, 2, 3], dtype=object)]
print(enc.categories_)
# ['gender_Female' 'gender_Male' 'group_1' 'group_2' 'group_3']
print(enc.get_feature_names(['gender', 'group']))

inverse_transform

from sklearn.preprocessing import OneHotEncoder
enc = OneHotEncoder(handle_unknown='ignore')
x_train = [['Male', 1], ['Female', 3], ['Female', 2]]
enc.fit(x_train)
x_test = enc.inverse_transform([[0, 1, 1, 0, 0], [0, 0, 0, 1, 0]])
# [['Male' 1]
#  [None 2]]
print(x_test)

LabelEncoder

• classes_
• inverse_transform

from sklearn.preprocessing import LabelEncoder

x_train = ["paris", "paris", "tokyo", "amsterdam"]
x_test = ["tokyo", "tokyo", "paris"]
le = LabelEncoder()
le.fit(x_train)
# ['amsterdam' 'paris' 'tokyo']
print(le.classes_)
x_test = le.transform(x_test)
# [2 2 1]
print(x_test)
# ['tokyo' 'tokyo' 'paris']
print(le.inverse_transform([2,2,1]))