我有一个包含多个子查询和条件的查询，以建立各自具有自己分组的计算列。
问题是，某项未正确加入或未正确分组，我无法在所有行中获得正确的值。我选择了过去90天的原始行，但我的计算仅基于过去30天的记录。
SELECT
t.dealer,
t.rep,
t.group,
t.dtl1,
t.dtl2,
t.quant,
temp.sales as V1,
tmng.sales as V2,
tall.sales as V3
FROM
products t
left outer JOIN (
SELECT dealer,group,dtl1,dtl2, SUM(cast(quant as int)) as sales
FROM products
WHERE last_ship_date > CURRENT_DATE - 30 DAYS
GROUP BY dealer,group,dtl1,dtl2
) temp ON temp.dealer = t.dealer and temp.group = t.group and temp.dtl1 = t.dtl1 and temp.dtl2 = t.dtl2
left outer JOIN (
SELECT sales_rep, group,dtl1,dtl2, SUM(cast(quant as int)) as sales
FROM products
WHERE last_ship_date > CURRENT_DATE - 30 DAYS
GROUP BY sales_rep,group,dtl1,dtl2
) tmng ON tmng.sales_rep = t.sales_rep AND tmng.group = t.group AND tmng.dtl1 = t.dtl1 and tmng.dtl2 = t.dtl2
left outer JOIN (
SELECT dealer, sales_rep, group,dtl1,dtl2,SUM(cast(quant as int)) as sales
FROM products
WHERE last_ship_date > CURRENT_DATE - 30 DAYS
group by dealer,sales_rep, group,dtl1,dtl2
) tall on tall.dealer = t.dealer and tall.sales_rep = t.sales_rep and tall.group = t.group and tall.dtl1 = t.dtl1 and tall.dtl2 = t.dtl2
WHERE t.last_ship_date > CURRENT_DATE - 90 DAYS;
这个想法是，一个产品在Group / Dtl1 / Dtl2中是唯一的。因此，如果我查询单个产品（具有该唯一组合），则可以得到每行的特定总数量，但是我需要解决的是我的最后三列，其中我需要最近30天的“数量”总和经销商，代表和总数。
现在，我得到的行数正确，但是最后三列中的行值和空值不正确。
Dealer | Rep | Group | Dtl1 | Dtl2 | qty | V1 | V2 | V3
========================================================================
21177 | 165 | 4867 | 2878 | 29 | 1 | null | null | null
19445 | 182 | 4867 | 2878 | 29 | 1 | null | null | null
20949 | 340 | 4867 | 2878 | 29 | 2 | null | null | null
21347 | 353 | 4867 | 2878 | 29 | 1 | null | null | null
23068 | 353 | 4867 | 2878 | 29 | 1 | null | null | null
17195 | 353 | 4867 | 2878 | 29 | 6 | null | null | null
23040 | 353 | 4867 | 2878 | 29 | 1 | null | null | null
5620 | 380 | 4867 | 2878 | 29 | 4 | null | null | null
23009 | 380 | 4867 | 2878 | 29 | 3 | null | null | null
13529 | 390 | 4867 | 2878 | 29 | 1 | null | null | null
22706 | 394 | 4867 | 2878 | 29 | 1 | 1 | 1 | 1
17049 | 394 | 4867 | 2878 | 29 | 4 | 1 | null | null
17879 | 397 | 4867 | 2878 | 29 | 1 | null | null | null
21320 | 445 | 4867 | 2878 | 29 | 1 | null | null | null
21231 | 457 | 4867 | 2878 | 29 | 1 | 1 | 1 | 1
我想要得到的是这个
Dealer | Rep | Group | Dtl1 | Dtl2 | qty | V1 | V2 | V3
================================================================
21177 | 165 | 4867 | 2878 | 29 | 1 | 1 | 1 | 29
19445 | 182 | 4867 | 2878 | 29 | 1 | 1 | 1 | 29
20949 | 340 | 4867 | 2878 | 29 | 2 | 2 | 2 | 29
21347 | 353 | 4867 | 2878 | 29 | 1 | 1 | 9 | 29
23068 | 353 | 4867 | 2878 | 29 | 1 | 1 | 9 | 29
17195 | 353 | 4867 | 2878 | 29 | 6 | 6 | 9 | 29
23040 | 353 | 4867 | 2878 | 29 | 1 | 1 | 9 | 29
5620 | 380 | 4867 | 2878 | 29 | 4 | 4 | 7 | 29
23009 | 380 | 4867 | 2878 | 29 | 3 | 3 | 7 | 29
13529 | 390 | 4867 | 2878 | 29 | 1 | 1 | 1 | 29
22706 | 394 | 4867 | 2878 | 29 | 1 | 1 | 5 | 29
17049 | 394 | 4867 | 2878 | 29 | 4 | 4 | 5 | 29
17879 | 397 | 4867 | 2878 | 29 | 1 | 1 | 1 | 29
21320 | 445 | 4867 | 2878 | 29 | 1 | 1 | 1 | 29
21231 | 457 | 4867 | 2878 | 29 | 1 | 1 | 1 | 29
因此，V1是最近30天的Dealer / group / dtl1 / dtl2的数量之和，V2是最近30天的Rep / Group / Dtl1 / Dtl2的数量之和，V3是最近30天该Group / Dtl1 / Dtl2的总计（数量）。
我为什么错了？
答案 0 :(得分：1)
尝试一下：
WITH T (DEALER, REP, GROUP, DTL1, Dtl2, QUANT, LAST_SHIP_DATE) AS (
VALUES (21177, 165, 4867, 2878, 29, 1, CURRENT DATE - 10 DAY),
(19445, 182, 4867, 2878, 29, 1, CURRENT DATE - 10 DAY),
(20949, 340, 4867, 2878, 29, 2, CURRENT DATE - 10 DAY),
(21347, 353, 4867, 2878, 29, 1, CURRENT DATE - 10 DAY),
(23068, 353, 4867, 2878, 29, 1, CURRENT DATE - 10 DAY),
(17195, 353, 4867, 2878, 29, 6, CURRENT DATE - 10 DAY),
(23040, 353, 4867, 2878, 29, 1, CURRENT DATE - 10 DAY),
(5620, 380, 4867, 2878, 29, 4, CURRENT DATE - 10 DAY),
(23009, 380, 4867, 2878, 29, 3, CURRENT DATE - 10 DAY),
(13529, 390, 4867, 2878, 29, 1, CURRENT DATE - 10 DAY),
(22706, 394, 4867, 2878, 29, 1, CURRENT DATE - 10 DAY),
(17049, 394, 4867, 2878, 29, 4, CURRENT DATE - 10 DAY),
(17879, 397, 4867, 2878, 29, 1, CURRENT DATE - 10 DAY),
(21320, 445, 4867, 2878, 29, 1, CURRENT DATE - 10 DAY),
(21231, 457, 4867, 2878, 29, 1, CURRENT DATE - 10 DAY)
, (1, 1, 4867, 2878, 29, 1, CURRENT DATE - 100 DAY)
)
SELECT T.DEALER, T.REP, T.GROUP, T.DTL1, T.Dtl2, T.QUANT
/*
, SUM(CASE WHEN last_ship_date > CURRENT_DATE - 30 DAYS THEN QUANT END) OVER(PARTITION BY DEALER, GROUP, DTL1, Dtl2) V1
, SUM(CASE WHEN last_ship_date > CURRENT_DATE - 30 DAYS THEN QUANT END) OVER(PARTITION BY REP, GROUP, DTL1, Dtl2) V2
, SUM(CASE WHEN last_ship_date > CURRENT_DATE - 30 DAYS THEN QUANT END) OVER(PARTITION BY GROUP, DTL1, Dtl2) V3
FROM T;
*/
, COALESCE(T1.V1, 0) V1, COALESCE(T2.V2, 0) V2, COALESCE(T3.V3, 0) V3
FROM T
LEFT JOIN (
SELECT DEALER, GROUP, DTL1, Dtl2, SUM(QUANT) V1
FROM T
WHERE last_ship_date > CURRENT_DATE - 30 DAYS
GROUP BY DEALER, GROUP, DTL1, Dtl2
) T1 ON T1.DEALER=T.DEALER AND T1.GROUP=T.GROUP AND T1.DTL1=T.DTL1 AND T1.DTL2=T.DTL2
LEFT JOIN (
SELECT REP, GROUP, DTL1, Dtl2, SUM(QUANT) V2
FROM T
WHERE last_ship_date > CURRENT_DATE - 30 DAYS
GROUP BY REP, GROUP, DTL1, Dtl2
) T2 ON T2.REP=T.REP AND T2.GROUP=T.GROUP AND T2.DTL1=T.DTL1 AND T2.DTL2=T.DTL2
LEFT JOIN (
SELECT GROUP, DTL1, Dtl2, SUM(QUANT) V3
FROM T
WHERE last_ship_date > CURRENT_DATE - 30 DAYS
GROUP BY GROUP, DTL1, Dtl2
) T3 ON T3.GROUP=T.GROUP AND T3.DTL1=T.DTL1 AND T3.DTL2=T.DTL2;