我尝试根据O_Diagnosis操作生成一个组号。我想将当前和下一个O_Diagnosis操作之间的操作分组。
问题是这些数据没有任何支持分组的列。所以我需要几乎从头开始生成这个列。我想要实现的目标如下：
Site SFC Date Time Operation Times_processed Diagnosis_group_nr
P500 B218YW 2017-03-21 10:16:23 O_SF_WRAP 1 1
P500 B218YW 2017-03-23 06:07:53 O_SF_WRAP_CURE 1 1
P500 B218YW 2017-03-23 14:23:41 O_DIAGNOSIS 1 2
P500 B218YW 2017-03-28 10:07:55 O_SF_WRAP 2 2
P500 B218YW 2017-03-28 15:02:21 O_SF_WRAP_CURE 2 2
P500 B218YW 2017-03-29 20:38:06 O_DIAGNOSIS 2 3
P500 B218YW 2017-03-29 20:39:15 O_DIAGNOSIS 3 4
P500 B218YW 2017-04-27 08:43:53 O_SF_WRAP 3 4
P500 B218YW 2017-04-27 12:43:25 O_SF_WRAP_CURE 3 4
P500 B218YW 2017-04-27 13:05:16 O_SF_PCA_ASSEMBLY 1 4
P500 B218YW 2017-04-27 13:19:15 O_DIAGNOSIS 4 5
P500 B218YW 2017-04-27 15:23:09 O_SF_PCA_ASSEMBLY 2 5
P500 B218YW 2017-04-27 16:10:32 O_SF_ELEC_INT_TEST 1 1
P500 B218YW 2017-04-27 22:16:23 O_RECORD_REVIEW 1 1
The Times处理列计算SFC执行操作的频率。这就是为什么你看到1为O_SF_WRAP的例子，当下次从O_Diagnosis发回SFC时，这将变为2。
是否可以创建一个函数或SQL视图来实现这一目标？我正在使用SQL Server 2012.
到目前为止，对于同一个证监会来说，我目前所拥有的是：
Site SFC Date Time Operation Times_processed Diagnosis_group_nr
P500 B218YW 2017-03-21 10:16:23 O_SF_WRAP 1 1
P500 B218YW 2017-03-23 06:07:53 O_SF_WRAP_CURE 1 1
P500 B218YW 2017-03-23 14:23:41 O_DIAGNOSIS 1 2
P500 B218YW 2017-03-28 10:07:55 O_SF_WRAP 2 2
P500 B218YW 2017-03-28 15:02:21 O_SF_WRAP_CURE 2 2
P500 B218YW 2017-03-29 20:38:06 O_DIAGNOSIS 2 3
P500 B218YW 2017-03-29 20:39:15 O_DIAGNOSIS 3 3
P500 B218YW 2017-04-27 08:43:53 O_SF_WRAP 3 3
P500 B218YW 2017-04-27 12:43:25 O_SF_WRAP_CURE 3 3
P500 B218YW 2017-04-27 13:05:16 O_SF_PCA_ASSEMBLY 1 4
P500 B218YW 2017-04-27 13:19:15 O_DIAGNOSIS 4 4
P500 B218YW 2017-04-27 15:23:09 O_SF_PCA_ASSEMBLY 2 2
P500 B218YW 2017-04-27 16:10:32 O_SF_ELEC_INT_TEST 1 1
P500 B218YW 2017-04-27 22:16:23 O_RECORD_REVIEW 1 1
到目前为止，这是我的代码：
WITH
cteNextRow
AS
(
SELECT
opl.[PL SITE]
, opl.[PL SFC]
, opl.[Pl End Time Local]
, opl.[PL OPERATION]
, LEAD(opl.[PL OPERATION], 1) OVER (PARTITION BY opl.[PL SFC] ORDER BY opl.[Pl End Time Local]) AS NEXT_OPERATION
, LEAD(opl.[Pl Times Processed], 1) OVER (PARTITION BY opl.[PL SFC] ORDER BY opl.[Pl End Time Local]) AS NEXT_TIMES_PROCESSED
, max(opl.[Pl Times Processed]) OVER (PARTITION BY opl.[PL SFC] ORDER BY opl.[PL SFC], opl.[Pl End Time Local]
ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) TOTAL_TIMES_PROCESSED
,LAST_VALUE(opl.[PL OPERATION]) OVER (PARTITION BY opl.[PL SFC] ORDER BY opl.[PL SFC], opl.[Pl End Time Local]
ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING
) as LAST_OPERATION
,case when LAST_VALUE(opl.[PL OPERATION]) OVER (PARTITION BY opl.[PL SFC] ORDER BY opl.[PL SFC], opl.[Pl End Time Local]
ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) not like '%review%'
then 0
when LAST_VALUE(opl.[PL OPERATION]) OVER (PARTITION BY opl.[PL SFC] ORDER BY opl.[PL SFC], opl.[Pl End Time Local]
ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) like '%review%' and max(opl.[Pl Times Processed]) OVER (PARTITION BY opl.[PL SFC] ORDER BY opl.[PL SFC], opl.[Pl End Time Local]
ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) = 1
then 1
else 9999
end as REVIEW_IND
FROM Report1 opl
where [pl sfc] = 'B218YW'
)
select
opl.[pl Site] AS PROCESS_SITE
, opl.[PL SFC] AS PROCESS_SFC
, opl.[Pl End Time Local] PROCESS_DATE_TIME
, opl.[PL OPERATION] AS PROCESS_OPERATION
, cte.NEXT_OPERATION AS PROCESS_NEXT_OPERATION
, opl.[Pl Times Processed] AS PROCESS_TIMES_PROCESSED
, CASE
when cte.REVIEW_IND = 9999 and cte.NEXT_TIMES_PROCESSED = opl.[Pl Times Processed] and cte.NEXT_OPERATION like '%DIAG%'
then LEAD(opl.[Pl Times Processed], 1) OVER (PARTITION BY opl.[PL SFC] ORDER BY opl.[Pl End Time Local])
when cte.REVIEW_IND = 9999 and cte.NEXT_TIMES_PROCESSED < opl.[Pl Times Processed]
then opl.[Pl Times Processed]
when cte.REVIEW_IND = 9999 and NEXT_TIMES_PROCESSED is null
then opl.[Pl Times Processed]
when cte.REVIEW_IND = 9999 and cte.NEXT_TIMES_PROCESSED is not null
then cte.NEXT_TIMES_PROCESSED
else REVIEW_IND
END AS Diagnosis_group_nr
FROM Report1 opl
join cteNextRow cte on cte.[PL SITE] = opl.[pl Site] and cte.[PL sfc] = opl.[PL SFC] and cte.[pl operation] = opl.[pl OPERATION] and opl.[Pl End Time Local] = cte.[Pl End Time Local]
答案 0 :(得分：0)
我不确定为什么您的分组编号会在所需输出结尾处重置为1，但如果这是拼写错误，您可以使用简单的窗口条件row_number和sum得到Times_processed和`Diagnosis_group_nr＆＃39;的数量你需要分别：
declare @t table(Site nvarchar(50),SFC nvarchar(50),DateValue datetime,Operation nvarchar(50));
insert into @t values
('P500','B218YW','2017-03-21 10:16:23','O_SF_WRAP')
,('P500','B218YW','2017-03-23 06:07:53','O_SF_WRAP_CURE')
,('P500','B218YW','2017-03-23 14:23:41','O_DIAGNOSIS')
,('P500','B218YW','2017-03-28 10:07:55','O_SF_WRAP')
,('P500','B218YW','2017-03-28 15:02:21','O_SF_WRAP_CURE')
,('P500','B218YW','2017-03-29 20:38:06','O_DIAGNOSIS')
,('P500','B218YW','2017-03-29 20:39:15','O_DIAGNOSIS')
,('P500','B218YW','2017-04-27 08:43:53','O_SF_WRAP')
,('P500','B218YW','2017-04-27 12:43:25','O_SF_WRAP_CURE')
,('P500','B218YW','2017-04-27 13:05:16','O_SF_PCA_ASSEMBLY')
,('P500','B218YW','2017-04-27 13:19:15','O_DIAGNOSIS')
,('P500','B218YW','2017-04-27 15:23:09','O_SF_PCA_ASSEMBLY')
,('P500','B218YW','2017-04-27 16:10:32','O_SF_ELEC_INT_TEST')
,('P500','B218YW','2017-04-27 22:16:23','O_RECORD_REVIEW');
select *
,row_number() over (partition by Operation order by DateValue) as Times_processed
,sum(case when Operation = 'O_DIAGNOSIS' then 1 else 0 end) over (order by DateValue)+1 as Diagnosis_group_nr
from @t
order by DateValue;
输出：
+------+--------+-------------------------+--------------------+-----------------+--------------------+
| Site | SFC | DateValue | Operation | Times_processed | Diagnosis_group_nr |
+------+--------+-------------------------+--------------------+-----------------+--------------------+
| P500 | B218YW | 2017-03-21 10:16:23.000 | O_SF_WRAP | 1 | 1 |
| P500 | B218YW | 2017-03-23 06:07:53.000 | O_SF_WRAP_CURE | 1 | 1 |
| P500 | B218YW | 2017-03-23 14:23:41.000 | O_DIAGNOSIS | 1 | 2 |
| P500 | B218YW | 2017-03-28 10:07:55.000 | O_SF_WRAP | 2 | 2 |
| P500 | B218YW | 2017-03-28 15:02:21.000 | O_SF_WRAP_CURE | 2 | 2 |
| P500 | B218YW | 2017-03-29 20:38:06.000 | O_DIAGNOSIS | 2 | 3 |
| P500 | B218YW | 2017-03-29 20:39:15.000 | O_DIAGNOSIS | 3 | 4 |
| P500 | B218YW | 2017-04-27 08:43:53.000 | O_SF_WRAP | 3 | 4 |
| P500 | B218YW | 2017-04-27 12:43:25.000 | O_SF_WRAP_CURE | 3 | 4 |
| P500 | B218YW | 2017-04-27 13:05:16.000 | O_SF_PCA_ASSEMBLY | 1 | 4 |
| P500 | B218YW | 2017-04-27 13:19:15.000 | O_DIAGNOSIS | 4 | 5 |
| P500 | B218YW | 2017-04-27 15:23:09.000 | O_SF_PCA_ASSEMBLY | 2 | 5 |
| P500 | B218YW | 2017-04-27 16:10:32.000 | O_SF_ELEC_INT_TEST | 1 | 5 |
| P500 | B218YW | 2017-04-27 22:16:23.000 | O_RECORD_REVIEW | 1 | 5 |
+------+--------+-------------------------+--------------------+-----------------+--------------------+
如果您想在多个Site和SFC上运行上述脚本，只需将这些列添加到两个窗口函数的partition by上，它们就可以正确地计算出来整个数据集一气呵成。