Pandas - 遍历表以获取累积天数和活动计数答案

【问题标题】：Pandas - Traverse Table to Get Accumulated Days and Activity CountPandas - 遍历表以获取累积天数和活动计数
【发布时间】：2022-08-18 17:58:51
【问题描述】：

我想在完成销售之前获得间隔天数、累积天数和非销售活动总数。

非常感谢任何帮助，因为我想不出一个干净的方法来做到这一点。我尝试使用 pandas\' shift 但我无法得到正确的结果。

输入熊猫表：

ID	Date	Activity
ID123	2021-01-01	Activity Type 1
ID123	2021-01-05	Activity Type 2
ID123	2021-01-07	Activity Type 1
ID123	2021-01-10	Sales
ID456	2021-01-01	Sales
ID456	2021-01-05	Sales
ID789	2021-01-20	Activity Type 1
ID789	2021-01-25	Sales
ID789	2021-01-27	Activity Type 2
ID789	2021-01-30	Sales

期望的输出：

ID	Date	Activity	Gap from Last Activity	Accumulated Days	Activity Count
ID123	2021-01-01	Activity Type 1			1
ID123	2021-01-05	Activity Type 2	5	5	2
ID123	2021-01-07	Activity Type 1	2	7	3
ID123	2021-01-10	Sales	3	10	3
ID456	2021-01-01	Sales			0
ID456	2021-01-05	Sales			0
ID789	2021-01-20	Activity Type 1			1
ID789	2021-01-25	Sales	5	5	1
ID789	2021-01-27	Activity Type 2			1
ID789	2021-01-30	Sales	3	3	1

标签： python pandas

【解决方案1】：

IIUC，您可以使用groupby 操作：

df['Date'] = pd.to_datetime(df['Date'])
m = df['Activity'].str.startswith('Activity')
g = df.groupby(['ID', (~m[::-1]).cumsum()])

df2 = df.assign(**{'Gap': g['Date'].diff(),
                   'Accumulated': lambda d: d.groupby('ID')['Gap']
                                             .apply(lambda x: x.fillna('0').cumsum()
                                                               .mask(x.isna())
                                                   ),
                   'Count': lambda d: g.cumcount().mask(d['Gap'].isna()).ffill(limit=1)
                   })

输出：

      ID       Date         Activity    Gap Accumulated  Count
0  ID123 2021-01-01  Activity Type 1    NaT         NaT    NaN
1  ID123 2021-01-05  Activity Type 2 4 days      4 days    1.0
2  ID123 2021-01-07  Activity Type 1 2 days      6 days    2.0
3  ID123 2021-01-10            Sales 3 days      9 days    3.0
4  ID456 2021-01-01            Sales    NaT         NaT    3.0
5  ID456 2021-01-05            Sales    NaT         NaT    NaN
6  ID789 2021-01-20  Activity Type 1    NaT         NaT    NaN
7  ID789 2021-01-25            Sales 5 days      5 days    1.0
8  ID789 2021-01-27  Activity Type 2    NaT         NaT    1.0
9  ID789 2021-01-30            Sales 3 days      8 days    1.0

【讨论】：