我正在尝试在replace
duplicate
中的多个columns
上pandas
df
个值。对于下面的df
,我有相应的日期和值。每个日期的所有值均相同。我只想保留每个日期的第一个值,并用duplicate
替换以下np.nan
值。以下是我的尝试:
import pandas as pd
import numpy as np
d = ({
'Date' : ['1/1/18','1/1/18','1/1/18','2/1/18','2/1/18','3/2/18','3/2/18','3/2/18'],
'Val_D' : [10,10,10,22,22,10,10,10],
'Val_M' : [100,100,100,100,100,240,240,240],
})
df = pd.DataFrame(data = d)
df['Date'] = pd.to_datetime(df['Date'], format= '%d/%m/%y')
pd.Series([10,22,100,240]).duplicated()
dup = df.apply(pd.Series.duplicated, axis = 1)
df = df.where(~dup,np.nan)
print(df)
预期输出:
Date Val_D Val_M
0 1/1/18 10 100
1 1/1/18
2 1/1/18
3 2/1/18 22
4 2/1/18
5 3/2/18 10 240
6 3/2/18
7 3/2/18
答案 0 :(得分:2)
好吧,一种方法是简单地使用diff
+ ne
s = df[['Val_D', 'Val_M']]
df[['Val_D', 'Val_M']] = s[s.diff().ne(0)].fillna('')
即使这会产生预期的输出,列的dtypes
也会变成object
,并且您会失去数字的矢量化能力。因此,我建议您不要这样做。没有最后一个fillna('')
,您将获得
Date Val_D Val_M
0 2018-01-01 10.0 100.0
1 2018-01-01 NaN NaN
2 2018-01-01 NaN NaN
3 2018-01-02 22.0 NaN
4 2018-01-02 NaN NaN
5 2018-02-03 10.0 240.0
6 2018-02-03 NaN NaN
7 2018-02-03 NaN NaN
dtypes
是float
。现在,有了fillna('')
件,您就可以
Date Val_D Val_M
0 2018-01-01 10 100
1 2018-01-01
2 2018-01-01
3 2018-01-02 22
4 2018-01-02
5 2018-02-03 10 240
6 2018-02-03
7 2018-02-03
具有object
dtypes。
答案 1 :(得分:1)
我将where
与diff
一起使用
df[['Val_D', 'Val_M']].where(df[['Val_D', 'Val_M']].diff().ne(0))
Out[1136]:
Val_D Val_M
0 10.0 100.0
1 NaN NaN
2 NaN NaN
3 22.0 NaN
4 NaN NaN
5 10.0 240.0
6 NaN NaN
7 NaN NaN
df[['Val_D', 'Val_M']]= df[['Val_D', 'Val_M']].where(df[['Val_D', 'Val_M']].diff().ne(0))