这是我的样本数据
shop_code product_id doc_date ship_count mark_1 outputer y
-----------------------------------------------------------------------------
1 00664НСК 11628 2015-01-03 00:00:00.000 12 1 8 1
2 00664НСК 11628 2015-01-05 00:00:00.000 7 1 8 1
3 00664НСК 11628 2015-01-06 00:00:00.000 24 0 8 1
4 00664НСК 11628 2015-01-07 00:00:00.000 18 1 8 1
5 00664НСК 11628 2015-01-08 00:00:00.000 12 1 8 1
6 00664НСК 11628 2015-01-09 00:00:00.000 18 0 8 1
7 00664НСК 11628 2015-01-10 00:00:00.000 6 0 6 1
8 00664НСК 11628 2015-01-11 00:00:00.000 6 1 6 1
9 00664НСК 11628 2015-01-12 00:00:00.000 6 1 6 1
10 00664НСК 11628 2015-01-13 00:00:00.000 18 1 12 0
11 00664НСК 11628 2015-01-14 00:00:00.000 6 1 6 0
12 00664НСК 11628 2015-01-15 00:00:00.000 18 1 12 0
13 00664НСК 11628 2015-01-16 00:00:00.000 12 1 12 1
14 00664НСК 11628 2015-01-17 00:00:00.000 18 1 12 1
15 00664НСК 11628 2015-01-18 00:00:00.000 12 1 12 1
16 00664НСК 11628 2015-01-19 00:00:00.000 10 1 10 0
17 00664НСК 11628 2015-01-20 00:00:00.000 24 1 12 0
18 00664НСК 11628 2015-01-21 00:00:00.000 6 1 6 0
19 00664НСК 11628 2015-01-24 00:00:00.000 6 1 6 0
20 00664НСК 11628 2015-01-25 00:00:00.000 6 0 6 0
21 00664НСК 11628 2015-01-26 00:00:00.000 10 0 10 1
22 00664НСК 11628 2015-01-27 00:00:00.000 6 1 6 0
23 00664НСК 11628 2015-01-28 00:00:00.000 10 1 10 0
24 00664НСК 11628 2015-01-29 00:00:00.000 70 0 12 1
25 00664НСК 11628 2015-01-30 00:00:00.000 100 1 12 1
类似的问题，我问过R并得到了可行的解决方案，但现在我想使用T-SQL来完成。
我需要观察这样的条件：如果y = 1且mark1 = 1，则必须将mark1 = 1的输出替换为输出变量中y = 0且mark1 = 1的第一个值。
如果输出中第一个用于Y = 0且mark1 = 1的值大于ship_count，则在输出中保留ship_count的实际值
用于输出的mark1的零类别，我们不会碰。
此操作必须由ship_code + product_id组完成
因此所需的输出应如下所示：
shop_code product_id doc_date ship_count mark_1 outputer y
----------------------------------------------------------------------------
1 00664НСК 11628 2015-01-03 00:00:00.000 12 1 *12 1
2 00664НСК 11628 2015-01-05 00:00:00.000 7 1 *7 1
3 00664НСК 11628 2015-01-06 00:00:00.000 24 0 24 1
4 00664НСК 11628 2015-01-07 00:00:00.000 18 1 *12 1
5 00664НСК 11628 2015-01-08 00:00:00.000 12 1 *12 1
6 00664НСК 11628 2015-01-09 00:00:00.000 18 0 18 1
7 00664НСК 11628 2015-01-10 00:00:00.000 6 0 6 1
8 00664НСК 11628 2015-01-11 00:00:00.000 6 1 6 1
9 00664НСК 11628 2015-01-12 00:00:00.000 6 1 6 1
10 00664НСК 11628 2015-01-13 00:00:00.000 18 1 *12 0
11 00664НСК 11628 2015-01-14 00:00:00.000 6 1 6 0
12 00664НСК 11628 2015-01-15 00:00:00.000 18 1 12 0
13 00664НСК 11628 2015-01-16 00:00:00.000 12 1 *10 1
14 00664НСК 11628 2015-01-17 00:00:00.000 18 1 *10 1
15 00664НСК 11628 2015-01-18 00:00:00.000 12 1 *10 1
16 00664НСК 11628 2015-01-19 00:00:00.000 10 1 10 0
17 00664НСК 11628 2015-01-20 00:00:00.000 24 1 12 0
18 00664НСК 11628 2015-01-21 00:00:00.000 6 1 6 0
19 00664НСК 11628 2015-01-24 00:00:00.000 6 1 6 0
20 00664НСК 11628 2015-01-25 00:00:00.000 6 0 6 0
21 00664НСК 11628 2015-01-26 00:00:00.000 10 0 10 1
22 00664НСК 11628 2015-01-27 00:00:00.000 6 1 6 1
23 00664НСК 11628 2015-01-28 00:00:00.000 20 1 *12 0
24 00664НСК 11628 2015-01-29 00:00:00.000 70 1 12 0
25 00664НСК 11628 2015-01-30 00:00:00.000 100 1 12 1
答案 0 :(得分：1)
晚上好
您应该使用案例声明来完成工作。
要查找描述子句的第一个值，请使用子查询，在该子查询中保留所需的顺序（order by）并选择前1个值。
尝试一下，如果您遇到一些问题，请再次询问。