Pandas：标记删除重复记录

Pandas提供了duplicated、Index.duplicated、drop_duplicates函数来标记及删除重复记录

duplicated函数用于标记Series中的值、DataFrame中的记录行是否是重复，重复为True，不重复为False

pandas.DataFrame.duplicated(self, subset=None, keep='first')
pandas.Series.duplicated(self, keep='first')

参数解释如下：
subset：用于识别重复的列标签或列标签序列，默认所有列标签

keep=‘frist’：除了第一次出现外，其余相同的被标记为重复

keep=‘last’：除了最后一次出现外，其余相同的被标记为重复

keep=False：所有相同的都被标记为重复

import  pandas as pd
import  numpy as np

#标记DataFrame重复例子
df = pd.DataFrame({'col1': ['one', 'one', 'two', 'two', 'two', 'three', 'four'], 'col2': [1, 2, 1, 2, 1, 1, 1],
                   'col3':['AA','BB','CC','DD','EE','FF','GG']},index=['a', 'a', 'b', 'c', 'b', 'a','c'])

'''
    col1  col2 col3
a    one     1   AA
a    one     2   BB
b    two     1   CC
c    two     2   DD
b    two     1   EE
a  three     1   FF
c   four     1   GG
'''

#duplicated(self, subset=None, keep='first')
#根据列名标记
#keep='first'
print(df.duplicated())#默认所有列，无重复记录
'''
a    False
a    False
b    False
c    False
b    False
a    False
c    False
dtype: bool
'''
print(df.duplicated('col1'))#第二、四、五行被标记为重复

'''
a    False
a     True
b    False
c     True
b     True
a    False
c    False
dtype: bool
'''
print(df.duplicated(['col1','col2']))#第五行被标记为重复
'''
a    False
a    False
b    False
c    False
b     True
a    False
c    False
dtype: bool
'''

#keep='last'
df.duplicated('col1','last')#第一、三、四行被标记重复
df.duplicated(['col1','col2'],keep='last')#第三行被标记为重复


#keep=False
df.duplicated('col1',False)#Series([True,True,True,True,True,False,False],index=['a','a','b','c','b','a','c'])
df.duplicated(['col1','col2'],keep=False)#在col1和col2列上出现相同的，都被标记为重复
type(df.duplicated(['col1','col2'],keep=False))#pandas.core.series.Series


print(df.index.duplicated())#默认keep='first',第二、五、七行被标记为重复
# [False  True False False  True  True  True]
print(df.index.duplicated(keep='last'))#第一、二、三、四被标记为重复
# [ True  True  True  True False False False]

print(df[df.index.duplicated()])#获取重复记录行
'''
    col1  col2 col3
a    one     2   BB
b    two     1   EE
a  three     1   FF
c   four     1   GG
'''


print(df[~df.index.duplicated('last')])#获取不重复记录行
'''
    col1  col2 col3
b    two     1   EE
a  three     1   FF
c   four     1   GG
'''

#标记Series重复例子
#duplicated(self, keep='first')
s = pd.Series(['one', 'one', 'two', 'two', 'two', 'three', 'four'] ,
              index= ['a', 'a', 'b', 'c', 'b', 'a','c'],
              name='sname')
s.duplicated()
s.duplicated('last')
s.duplicated(False)
#根据索引标记
s.index.duplicated()
s.index.duplicated('last')
s.index.duplicated(False)

drop_duplicates函数用于删除Series、DataFrame中重复记录，并返回删除重复后的结果

pandas.DataFrame.drop_duplicates(self, subset=None, keep='first', inplace=False)

pandas.Series.drop_duplicates(self, keep='first', inplace=False)

#删除DataFrame重复记录例子
#drop_duplicates(self, subset=None, keep='first', inplace=False)

print(df.drop_duplicates())
'''
a    one     1   AA
a    one     2   BB
b    two     1   CC
c    two     2   DD
b    two     1   EE
a  three     1   FF
c   four     1   GG
'''
print(df.drop_duplicates('col1'))#删除了df.duplicated('col1')标记的重复记录
'''
a    one     1   AA
b    two     1   CC
a  three     1   FF
c   four     1   GG
'''

print(df.drop_duplicates('col1','last'))#删除了df.duplicated('col1','last')标记的重复记录
'''
    col1  col2 col3
a    one     2   BB
b    two     1   EE
a  three     1   FF
c   four     1   GG
'''
df1.drop_duplicates(['col1','col2'])#删除了df.duplicated(['col1','col2'])标记的重复记录
df.drop_duplicates('col1',keep='last',inplace=True)#inplace=True表示在原DataFrame上执行删除操作
df.drop_duplicates('col1',keep='last',inplace=False)#inplace=False返回一个副本
#删除Series重复记录例子
#drop_duplicates(self, keep='first', inplace=False)
s.drop_duplicates()