Question

我有这个样本数据。

create table #student_info
(
 transaction_id int identity(1,1)
 ,parent_id int
 ,transaction_date date
 ,[status] varchar(1)
 ) 
insert into #student_info
(
 parent_id
 ,transaction_date
 ,status
)
values(1,'2017-01-01','F'),(1,'2017-01-02','F'),(1,'2017-01-03','F'),
      (1,'2017-01-04','F'),(1,'2017-01-05','F'),(1,'2017-01-06','T'),
      (1,'2017-01-07','F'),(1,'2017-01-08','F'),(1,'2017-01-14','F'),
      (1,'2017-01-15','F')
      ,(2,'2017-01-01','F'),(2,'2017-01-02','T'),(2,'2017-01-03','F'),
       (2,'2017-01-09','F'),(2,'2017-01-10','F'),(2,'2017-01-11','T'),
       (2,'2017-01-12','F'),(2,'2017-01-13','F')

      select * from #student_info

我需要使用以下逻辑填充另一个附加列（transaction_type），

parent_id

，直到第一个'T'状态，我应将transaction_type标记为＆＃39;购买＆＃39;
对于每个parent_id，在购买第一个transaction_type之后，购买日期的7天内所有连续的行都应标记为＆＃39;续订＆＃39;`< / p>

示例输出应附在下面的屏幕截图中。

感谢任何帮助。

Answer 1

查找状态为transaction_date的每个parent_id的第一个T，然后使用Case语句创建交易类型

这样的事情应该有所帮助

SELECT transaction_id,
       parent_id,
       transaction_date,
       status,
       transaction_type = CASE
                            WHEN transaction_date <= minT THEN 'Purchase'
                            WHEN Datediff(dd, minT, transaction_date) >= 7 THEN 'Renewal'
                            ELSE NULL
                          END
FROM   (SELECT *,
               minT = Min(CASE WHEN status = 'T' THEN transaction_date END)
                        OVER(partition BY parent_id) --first transaction_date for each parent id where status is T
        FROM   #student_info) a

结果：

+----------------+-----------+------------------+--------+------------------+
| transaction_id | parent_id | transaction_date | status | transaction_type |
+----------------+-----------+------------------+--------+------------------+
|              1 |         1 | 2017-01-01       | F      | Purchase         |
|              2 |         1 | 2017-01-02       | F      | Purchase         |
|              3 |         1 | 2017-01-03       | F      | Purchase         |
|              4 |         1 | 2017-01-04       | F      | Purchase         |
|              5 |         1 | 2017-01-05       | F      | Purchase         |
|              6 |         1 | 2017-01-06       | T      | Purchase         |
|              7 |         1 | 2017-01-07       | F      | NULL             |
|              8 |         1 | 2017-01-08       | F      | NULL             |
|              9 |         1 | 2017-01-14       | F      | Renewal          |
|             10 |         1 | 2017-01-15       | F      | Renewal          |
|             11 |         2 | 2017-01-01       | F      | Purchase         |
|             12 |         2 | 2017-01-02       | T      | Purchase         |
|             13 |         2 | 2017-01-03       | F      | NULL             |
|             14 |         2 | 2017-01-09       | F      | Renewal          |
|             15 |         2 | 2017-01-10       | F      | Renewal          |
|             16 |         2 | 2017-01-11       | T      | Renewal          |
|             17 |         2 | 2017-01-12       | F      | Renewal          |
|             18 |         2 | 2017-01-13       | F      | Renewal          |
+----------------+-----------+------------------+--------+------------------+

递归的前一行验证，用于分组数据集

1 个答案: