python - Pandas:Groupby，循环并添加一小时迭代组内的条件

Question

我有一个 df 如下，每个人都有一些重复的日期记录，我希望对其余数据保持完全相同的顺序/顺序，但只想为重复的日期记录添加一个小时。

df = pd.DataFrame({
    'name': ['Jim', 'Jim', 'Jim', 'Jim', 'Mike', 'Mike', 'Mike', 'Mike',
           'Polo', 'Polo', 'Polo', 'Polo', 'Polo', 'Tom', 'Tom', 'Tom', 'Tom'],
    'Item ID': [80, 68, 751, 'Started', 32, 126, 68,  'Started', 105, 68, 251, 76, 'Started', 82, 251, 23, "Started"],
    'Date':["2018-02-15", "2018-02-14", "2015-08-17", "2015-08-17",
            "2018-09-14", "2018-06-01", "2018-06-01", "2018-05-31",
            "2017-06-01", "2017-06-01", "2014-12-01", "2014-11-23", "2013-08-11",
            "2017-07-14", "2016-02-16", "2016-02-16", "2015-06-05"],
})

    name  Item ID        Date
0    Jim       80  2018-02-15
1    Jim       68  2018-02-14
2    Jim      751  2015-08-17 # duplicate date for Jim, add one hour here
3    Jim  Started  2015-08-17
4   Mike       32  2018-09-14
5   Mike      126  2018-06-01 # duplicate date for Mike, add one hour here
6   Mike       68  2018-06-01
7   Mike  Started  2018-05-31
8   Polo      105  2017-06-01 # duplicate date for Polo, add one hour here
9   Polo       68  2017-06-01
10  Polo      251  2014-12-01
11  Polo       76  2014-11-23
12  Polo  Started  2013-08-11
13   Tom       82  2017-07-14
14   Tom      251  2016-02-16 # duplicate date for Tom, add one hour here
15   Tom       23  2016-02-16
16   Tom  Started  2015-06-05

我写了一些代码但效果不佳而且效率很低。如果有人有任何想法请帮助，非常感谢。 我的预期结果:

    name  Item ID                 Date
0    Jim       80           2018-02-15
1    Jim       68           2018-02-14
2    Jim      751         2015-08-17 00:01:00 # added
3    Jim  Started           2015-08-17
4   Mike       32           2018-09-14
5   Mike      126         2018-06-01 00:01:00 # added
6   Mike       68           2018-06-01
7   Mike  Started           2018-05-31
8   Polo      105         2017-06-01 00:01:00 # added
9   Polo       68           2017-06-01
10  Polo      251           2014-12-01
11  Polo       76           2014-11-23
12  Polo  Started           2013-08-11
13   Tom       82           2017-07-14
14   Tom      251         2016-02-16 00:01:00 # added
15   Tom       23           2016-02-16
16   Tom  Started           2015-06-05

Answer 1

我们可以通过 iloc 以相反的顺序进行复制，然后将小时添加到原始日期

df.Date=pd.to_datetime(df.Date)+pd.to_timedelta(df.iloc[::-1].duplicated(['name','Date']).astype(int),unit='hour')
df
    name  Item ID                Date
0    Jim       80 2018-02-15 00:00:00
1    Jim       68 2018-02-14 00:00:00
2    Jim      751 2015-08-17 01:00:00
3    Jim  Started 2015-08-17 00:00:00
4   Mike       32 2018-09-14 00:00:00
5   Mike      126 2018-06-01 01:00:00
6   Mike       68 2018-06-01 00:00:00
7   Mike  Started 2018-05-31 00:00:00
8   Polo      105 2017-06-01 01:00:00
9   Polo       68 2017-06-01 00:00:00
10  Polo      251 2014-12-01 00:00:00
11  Polo       76 2014-11-23 00:00:00
12  Polo  Started 2013-08-11 00:00:00
13   Tom       82 2017-07-14 00:00:00
14   Tom      251 2016-02-16 01:00:00
15   Tom       23 2016-02-16 00:00:00
16   Tom  Started 2015-06-05 00:00:00