将时间序列转换为 pandas 中的间隔

Question

我有一个如下的时间序列：

    mode     timestamp               
0    A    2021-06-29 00:00:00-04:00
1    A    2021-06-29 01:00:00-04:00
2    A    2021-06-29 02:00:00-04:00
3    A    2021-06-29 03:00:00-04:00
4    B    2021-06-29 04:00:00-04:00
5    B    2021-06-29 05:00:00-04:00
6    B    2021-06-29 06:00:00-04:00
7    B    2021-06-29 07:00:00-04:00
8    B    2021-06-29 08:00:00-04:00
9    B    2021-06-29 09:00:00-04:00
10   B    2021-06-29 10:00:00-04:00
11   A    2021-06-29 11:00:00-04:00
12   A    2021-06-29 12:00:00-04:00
13   A    2021-06-29 13:00:00-04:00
14   A    2021-06-29 14:00:00-04:00
15   A    2021-06-29 15:00:00-04:00
16   A    2021-06-29 16:00:00-04:00
17   A    2021-06-29 17:00:00-04:00
18   A    2021-06-29 18:00:00-04:00
19   A    2021-06-29 19:00:00-04:00
20   C    2021-06-29 20:00:00-04:00
21   C    2021-06-29 21:00:00-04:00
22   C    2021-06-29 22:00:00-04:00
23   C    2021-06-29 23:00:00-04:00
24   C    2021-06-29 00:00:00-04:00
25   C    2021-06-29 01:00:00-04:00
26   C    2021-06-29 02:00:00-04:00
27   C    2021-06-29 03:00:00-04:00

我正在尝试生成一个字典列表，其中包含带有开始和结束时间的模式间隔。像这样的东西

[
{"mode": "A", start_time: "2021-06-29 00:00:00-04:00", end_time:"2021-06-29 03:00:00-04:00" },
{"mode": "B", start_time: "2021-06-29 04:00:00-04:00", end_time:"2021-06-29 10:00:00-04:00" },
{"mode": "A", start_time: "2021-06-29 11:00:00-04:00", end_time:"2021-06-29 19:00:00-04:00" },
{"mode": "C", start_time: "2021-06-29 20:00:00-04:00", end_time:"2021-06-29 03:00:00-04:00" },
]

时间序列非常长，循环起来似乎非常慢。关于如何实现这一目标有什么建议吗？

Answer 1

您可以使用自定义

groupby.agg

和

to_dict

:

out = (df.groupby('mode', as_index=False)
         .agg(**{'start_time': ('timestamp', 'first'),
                 'end_time': ('timestamp', 'last')})
         .to_dict('records')
      )

输出：

[{'mode': 'A', 'start_time': '2021-06-29 00:00:00-04:00', 'end_time': '2021-06-29 19:00:00-04:00'},
 {'mode': 'B', 'start_time': '2021-06-29 04:00:00-04:00', 'end_time': '2021-06-29 10:00:00-04:00'},
 {'mode': 'C', 'start_time': '2021-06-29 20:00:00-04:00', 'end_time': '2021-06-29 03:00:00-04:00'}]