作用:将分类型数据转换成连续的数值型变量。即是对不连续的数字或者文本进行编号。
import pandas as pd
#先创建一个数据框(包含缺失值)
df = pd.DataFrame({'auth':['spring','summer','fall','spring'],
'sply':['a','c','a','b'],
'name':['zhangsan','lisi','xiaohua','xiaomei']})
df
Out[124]:
auth sply name
0 spring a zhangsan
1 summer c lisi
2 fall a xiaohua
3 spring b xiaomei
categorical_name = ['auth','sply','name']
#定义一个循环函数,处理分类型特征,进行标签编码
def categorical_preprocessing(dataset,categorical_feature):
'''
param:
dataset:DataFrame,输入的数据集
categorical_feature:list,分类特征列名
'''
for feature in categorical_feature:
set_feature = set(dataset[feature])#将特征映射到集合中
dic_feature = {}
for i ,feat in enumerate(set_feature):
dic_feature[feat] = i
dataset[feature] = dataset[feature].map(dic_feature)
dataset = pd.get_dummies(dataset,columns=categorical_feature)
return dataset
#处理分类特征编码
dataset = categorical_preprocessing(df,categorical_name)
#分类变量编码结果
dataset
Out[74]:
auth_0 auth_1 auth_2 sply_0 ... name_0 name_1 name_2 name_3
0 0 1 0 0 ... 1 0 0 0
1 0 0 1 0 ... 0 0 1 0
2 1 0 0 0 ... 0 0 0 1
3 0 1 0 1 ... 0 1 0 0
补充:
标签编码完成后一般都需要再进行一次one-hot编码,变成只包含0和1的数据。
如果变量含有顺序,如:优、良、差。可以省略one-hot编码。