从熊猫数据框中获取前后日期时间

Dhiraj 发表于 Dev

迪拉吉

我有一个如下所示的数据框，显示时间、主题和 on_time。

    created_time       subject          on_time
2020-02-26 21:01:40      A        2020-02-26 21:08:40
2020-02-26 21:01:40      A        2020-02-26 21:01:43
2020-02-26 21:01:40      A        2020-02-26 20:50:55
2020-02-26 21:01:40      A        2020-02-26 21:44:40
2020-02-26 01:01:50      B        2020-02-26 01:01:52
2020-02-26 01:01:50      B        2020-02-26 00:08:40
2020-02-26 01:01:50      B        2020-02-26 01:08:40
2020-02-26 01:01:50      B        2020-02-26 00:59:15

我需要输出数据帧显示 created_time、subject 和 on_time 就在 created_time 之前和之后

    created_time       subject     on_time_preceding     on_time_following
2020-02-26 21:01:40      A        2020-02-26 21:44:40   2020-02-26 21:01:43
2020-02-26 01:01:50      B        2020-02-26 00:59:15   2020-02-26 01:01:52

on_time_preceding 是 created_time 之前最近的 on_time，on_time_following 是 created_time 之后最近的 on_time

尼加尼

有用

import pandas as pd
import io

table = """
created_time|subject|on_time
2020-02-26 21:01:40|A|2020-02-26 21:08:40
2020-02-26 21:01:40|A|2020-02-26 21:01:43
2020-02-26 21:01:40|A|2020-02-26 20:50:55
2020-02-26 21:01:40|A|2020-02-26 21:44:40
2020-02-26 01:01:50|B|2020-02-26 01:01:52
2020-02-26 01:01:50|B|2020-02-26 00:08:40
2020-02-26 01:01:50|B|2020-02-26 01:08:40
2020-02-26 01:01:50|B|2020-02-26 00:59:15
"""
df = pd.read_table(io.StringIO(table), parse_dates=['created_time', 'on_time'], sep='|')
print(df[df['created_time'] > df['on_time']]
      .sort_values('on_time')
      .drop_duplicates(['created_time', 'subject'], keep='last')
      .merge(df[df['created_time'] < df['on_time']].groupby('subject')['on_time'].min(),
             left_on='subject', right_index=True, suffixes=('_preceding', '_following'))
      .sort_values('subject')
      .reset_index(drop=True))

# output:     
         created_time subject   on_time_preceding   on_time_following
0 2020-02-26 21:01:40       A 2020-02-26 20:50:55 2020-02-26 21:01:43
1 2020-02-26 01:01:50       B 2020-02-26 00:59:15 2020-02-26 01:01:52

本文收集自互联网，转载请注明来源。

如有侵权，请联系 [email protected] 删除。