将 SQL Lag over PartitionBy 查询转换为 pandas

Question

我有一张包含一些客户购买数据的表格。我想知道顾客在店里的进出时间是多少，我为此编写了下面的 SQL 查询。如何将其转换为 python pandas？

SELECT MyTable.*, 
                   LAG(EventTypeID, 1, 0) 
                      OVER ( PARTITION BY ID,Name
                         ORDER BY Time) AS LastEvent , 
                   LEAD(EventTypeID, 1, 0) 
                      OVER ( PARTITION BY ID,Name
                         ORDER BY Time) AS NextEvent 
                FROM DL.dbo.DataTable MyTable

输入：

+-------------+--------+--------+-------+
| EventTypeID |   ID   |  Name  | Time  |
+-------------+--------+--------+-------+
|           1 | QWERTY | Joseph | 10.20 |
|           1 | QWERTY | Joseph | 10.25 |
+-------------+--------+--------+-------+

想要的结果：

+-------------+--------+--------+-------+-----------+-----------+
| EventTypeID |   ID   |  Name  | Time  | LastEvent | NextEvent |
+-------------+--------+--------+-------+-----------+-----------+
|      1      | QWERTY | Joseph | 10.20 |         0 |         1 |
|      1      | QWERTY | Joseph | 10.25 |         1 |         0 |
+-------------+--------+--------+-------+-----------+-----------+

Answer 1

df['LastEvent'] = df.sort_values(by=['Time'], ascending=True)\
                       .groupby(['ID','Name'])['EventTypeID'].shift(1)

df['NextEvent'] = df.sort_values(by=['Time'], ascending=True)\
                       .groupby(['ID','Name'])['EventTypeID'].shift(-1)

感谢Lev Gelman 的指导。上面的代码可以解决问题！