3-3 OneHot编码

3.3 OneHot编码

请参考《数据准备和特征工程》中的相关章节，调试如下代码。

---

基础知识

import pandas as pd

g = pd.DataFrame({"gender": ["man", 'woman', 'woman', 'man', 'woman']})
g

	gender
0	man
1	woman
2	woman
3	man
4	woman

# 将分类型特征转换为“虚拟变量”（哑变量）
pd.get_dummies(g)

	gender_man	gender_woman
0	1	0
1	0	1
2	0	1
3	1	0
4	0	1

df = pd.DataFrame({"gene_seg": ['A', 'B', 'B', 'A', 'A'],
                   'dis': ['gall', 'hyp', 'gall', 'hyp', 'hyp']
                  })
df

	gene_seg	dis
0	A	gall
1	B	hyp
2	B	gall
3	A	hyp
4	A	hyp

pd.get_dummies(df)

	gene_seg_A	gene_seg_B	dis_gall	dis_hyp
0	1	0	1	0
1	0	1	0	1
2	0	1	1	0
3	1	0	0	1
4	1	0	0	1

persons = pd.DataFrame({"name":["Newton", "Andrew Ng", "Jodan", "Bill Gates"],
                        'color':['white', 'yellow', 'black', 'white']})
persons

	name	color
0	Newton	white
1	Andrew Ng	yellow
2	Jodan	black
3	Bill Gates	white

# drop_first=True：去除冗余的特征
df_dum = pd.get_dummies(persons['color'], drop_first=True)   

# left_index、right_index：通过左侧/右侧的行索引进行合并
persons.merge(df_dum, left_index=True, right_index=True)

	name	color	white	yellow
0	Newton	white	1	0
1	Andrew Ng	yellow	0	1
2	Jodan	black	0	0
3	Bill Gates	white	1	0

from sklearn.preprocessing import OneHotEncoder

ohe = OneHotEncoder()
features = ohe.fit_transform(persons[['color']])
features.toarray()

array([[0., 1., 0.],
       [0., 0., 1.],
       [1., 0., 0.],
       [0., 1., 0.]])

# 去除冗余特征
features.toarray()[:, 1:]

array([[1., 0.],
       [0., 1.],
       [0., 0.],
       [1., 0.]])

项目案例

df = pd.DataFrame({
    "color": ['green', 'red', 'blue', 'red'],
    "size": ['M', 'L', 'XL', 'L'],
    "price": [29.9, 69.9, 99.9, 59.9],
    "classlabel": ['class1', 'class2', 'class1', 'class1']
})
df

	color	size	price	classlabel
0	green	M	29.9	class1
1	red	L	69.9	class2
2	blue	XL	99.9	class1
3	red	L	59.9	class1

size_mapping = {'XL': 3, 'L': 2, 'M': 1}

# map()方法相当于replace()方法，将特征“size”数值化
df['size'] = df['size'].map(size_mapping)    
df

	color	size	price	classlabel
0	green	1	29.9	class1
1	red	2	69.9	class2
2	blue	3	99.9	class1
3	red	2	59.9	class1

from sklearn.preprocessing import OneHotEncoder

# 获取OneHot编码模型ohe
ohe = OneHotEncoder()
fs = ohe.fit_transform(df[['color']])

# 去除冗余特征
fs_ohe = pd.DataFrame(fs.toarray()[:, 1:], columns=["color_green", 'color_red'])

# 将两表df, fs_ohe按列拼接为一个表
df = pd.concat([df, fs_ohe], axis=1)
df

	color	size	price	classlabel	color_green	color_red
0	green	1	29.9	class1	1.0	0.0
1	red	2	69.9	class2	0.0	1.0
2	blue	3	99.9	class1	0.0	0.0
3	red	2	59.9	class1	0.0	1.0

动手练习

cl = ohe.fit_transform(df[['classlabel']])

# 去除冗余特征
cl_ohe = pd.DataFrame(cl.toarray()[:, :-1], columns=["classlabel_class1"])

# 将两表df, cl_ohe按列拼接为一个表
df = pd.concat([df, cl_ohe], axis=1)
df

	color	size	price	classlabel	color_green	color_red	classlabel_class1
0	green	1	29.9	class1	1.0	0.0	1.0
1	red	2	69.9	class2	0.0	1.0	0.0
2	blue	3	99.9	class1	0.0	0.0	1.0
3	red	2	59.9	class1	0.0	1.0	1.0

pd.read_csv("/home/aistudio/data/data20513/breast-cancer.data").head()

	no-recurrence-events	30-39	premeno	30-34	0-2	no	3	left	left_low	no.1
0	no-recurrence-events	40-49	premeno	20-24	0-2	no	2	right	right_up	no
1	no-recurrence-events	40-49	premeno	20-24	0-2	no	2	left	left_low	no
2	no-recurrence-events	60-69	ge40	15-19	0-2	no	2	right	left_up	no
3	no-recurrence-events	40-49	premeno	0-4	0-2	no	2	right	right_low	no
4	no-recurrence-events	60-69	ge40	15-19	0-2	no	2	left	left_low	no

# 根据预览上面的csv文件之后，发现数据集没有特征行，因此标记为：header=None
df = pd.read_csv("/home/aistudio/data/data20513/breast-cancer.data", header=None).iloc[:, 1:]

print(df.shape)
df.head()

(286, 9)

	1	2	3	4	5	6	7	8	9
0	30-39	premeno	30-34	0-2	no	3	left	left_low	no
1	40-49	premeno	20-24	0-2	no	2	right	right_up	no
2	40-49	premeno	20-24	0-2	no	2	left	left_low	no
3	60-69	ge40	15-19	0-2	no	2	right	left_up	no
4	40-49	premeno	0-4	0-2	no	2	right	right_low	no

dataset = df.values
print(dataset)

# 前8列作为自变量X，最后1列作为样本标签，即因变量Y
X = dataset[:, 0:8]
X = X.astype(str)

Y = dataset[:, 8]

[['30-39' 'premeno' '30-34' ... 'left' 'left_low' 'no']
 ['40-49' 'premeno' '20-24' ... 'right' 'right_up' 'no']
 ['40-49' 'premeno' '20-24' ... 'left' 'left_low' 'no']
 ...
 ['60-69' 'ge40' '20-24' ... 'right' 'left_up' 'no']
 ['40-49' 'ge40' '30-34' ... 'left' 'left_low' 'no']
 ['50-59' 'ge40' '30-34' ... 'left' 'left_low' 'no']]

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

encoded_x = None
for i in range(0, X.shape[1]):

    label_encoder = LabelEncoder()   #只接受行向量
    feature = label_encoder.fit_transform(X[:,i])    # 将X的每1列转置为行向量进行数值化
    feature = feature.reshape(X.shape[0], 1)         #将feature转置为列向量

    onehot_encoder = OneHotEncoder(sparse=False)    # OneHot编码，只接受列向量
    feature = onehot_encoder.fit_transform(feature)

    if encoded_x is None:
        encoded_x = feature
    else:
        encoded_x = np.concatenate((encoded_x, feature), axis=1)

# 8列的特征通过变换后增大到41列
print("X shape: : ", encoded_x.shape)

X shape: :  (286, 41)

label_encoder = LabelEncoder()
feature = label_encoder.fit_transform(Y)

onehot_encoder = OneHotEncoder(sparse=False)    # OneHot编码，只接受列向量
feature = onehot_encoder.fit_transform(feature.reshape(Y.shape[0], 1))

print("Y shape: : ", feature.shape)

Y shape: :  (286, 2)