Pandas GroupBy日期范围取决于每一行

Question

我想做以下事情:

对于看起来像这样的数据框:

df = pd.DataFrame({"ID":["A", "A", "C" ,"B", "B"], "date":["06/24/2014","06/25/2014","06/23/2014","07/02/1999","07/02/1999"], "value": ["3","5","1","7","8"] })

我想按日期分组所有彼此在2天内的观察结果.然后,例如,前3行将被分组,最后两行将被分组.

到目前为止,我曾想过使用类似的东西:

df.groupby(df['date'].map(lambda x: x.month))

这种"模糊组合"的一般方法是什么？

谢谢,

Answer 1

您可以对行进行排序date,然后记录连续日期之间的差异.当差异大于2天时进行测试.取累计金额分配所需的组号:

import pandas as pd
df = pd.DataFrame({"ID":["A", "A", "C" ,"B", "B"], "date":["06/24/2014","06/25/2014","06/23/2014","07/02/1999","07/02/1999"], "value": ["3","5","1","7","8"] })
df['date'] = pd.to_datetime(df['date'])
df = df.sort_values(by='date')
df['group'] = (df['date'].diff() > pd.Timedelta(days=2)).cumsum()
print(df)

产量

  ID       date value  group
3  B 1999-07-02     7      0
4  B 1999-07-02     8      0
2  C 2014-06-23     1      1
0  A 2014-06-24     3      1
1  A 2014-06-25     5      1