离散特征值处理案例

import pandas as pd
import numpy as np
from sklearn.preprocessing import LabelEncoder, OneHotEncoder, LabelBinarizer, Binarizer

# 1.
data = pd.read_excel('score.xlsx')
print(data.head())

############ LabelEncoder:自动对元素进行编号，根据首字母大小顺序或者数字的大小顺序编号
le = LabelEncoder()
print(data['GENDER'])
gender = le.fit_transform(data['GENDER'])
data['GENDER_ID'] = gender
# del data['GENDER']
print(data)

################# OneHotEncoder: 独热编码,只有1位是1, 其余是0
ohe = OneHotEncoder(sparse=False);
print(data['GENDER_ID'])
gender = ohe.fit_transform(data[['GENDER_ID']])  # onehotencoder输入必须为二维矩阵,所以,要用[[ ]]
df_gender = pd.DataFrame(gender, columns=('ohe1', 'ohe2'))
data = pd.concat([data, df_gender], axis=1);
print(data)  # concat合并两个dataframe

# 或者用pd.get_dummies可以直接把字符串转成独热编码(onehotencoder不行)，相当于
# LabelEncoder+OneHotEncoder
a = pd.get_dummies(data['GENDER']);
print(a)
print(pd.concat([data, a], axis=1))

################ Binarizer: 以门槛值划分为0和1
data['SCORE'] = np.random.randint(1, 101, 6)
print(data)

bin = Binarizer(threshold=60)
pass_unpass = bin.fit_transform(data[['SCORE']])  # 输入必须为二维数据
data['pass/unpass'] = pass_unpass
print(data)

# 2.
data = pd.DataFrame({'name': ['joe', 'john', 'bob', 'kevin'],
                     'age': [10, 20, 30, 10],
                     'sex': ['male', 'male', 'female', 'male']})
data['age'] = LabelEncoder().fit_transform(data['age'])
print(data)

# LabelEncoder会自动对元素进行编号，根据首字母大小顺序或者数字的大小顺序编号。
data['name_LE'] = LabelEncoder().fit_transform(data['name'])
data['sex_LE'] = LabelEncoder().fit_transform(data['sex'])
print(data)
print(data.info())

# LabelBinarizer: 下标转化为0和1
data['sex_LB'] = LabelBinarizer().fit_transform(data['sex'])
print(data)
print(data.info())