的實際值的SQL查詢滿足預測某一週

Question

表：UnitsSold 列：

• 的ProductID
• 數量
• 日期
• ForecastOrActual - 可爲「A」的實際和「F」進行預測。

樣本數據是這樣的：

UnitsSold 
ProductID  Quantity Date   ForecastOrActual 
Widget   100   1-June-2015  F 
Widget   120   1-June-2015  A 
Gizmo   90   1-June-2015  F 
Gizmo   80   1-June-2015  A 
Doohickey  200   1-June-2015  A 

我找了一個SQL查詢，返回一個聚集指標指示如何實績滿足預測某一週。同時，任何的假設

感謝您的支持。由於

Answer 1

很難寫，因爲信息你在這裏的代碼，但想法是這樣的：

WITH 
    cte_A AS 
(SELECT qty,product, DATEPART(wk, DATE) AS WEEK 
    WHERE FORCAST='A' 
    GROUP BY product,DATEPART(wk, DATE)) 
, cte_F AS 
(SELECT qty,product, DATEPART(wk, DATE) AS WEEK 
    WHERE FORCAST='F' 
    GROUP BY product,DATEPART(wk, DATE)) 
SELECT cte_A.product,cte_A.week, 
CASE 
WHEN cte_F.qty - cte_A.qty >0 THEN 'Not met forcast' 
WHEN cte_F.qty - cte_A.qty <0 THEN 'Met forcast' 
END AS result 
FROM cte_A join cte_F 
ON cte_A.product=cte_F.product 
AND cte_A.week=cte_F.week 

如果你給我更多的信息，我也許能夠使此代碼更好地爲您。

日期部分功能將會給你一週的人數在一年的任何一天，你就可以使用這個數字每年如果您的數據多年跨越。 希望這有助於。

爲了更好的邏輯（-100〜100）性能，您可以將其更改爲：

WITH 
    cte_A AS 
(SELECT qty,product, DATEPART(wk, DATE) AS WEEK 
    WHERE FORCAST='A' 
    GROUP BY product,DATEPART(wk, DATE)) 
, cte_F AS 
(SELECT qty,product, DATEPART(wk, DATE) AS WEEK 
    WHERE FORCAST='F' 
    GROUP BY product,DATEPART(wk, DATE)) 
SELECT cte_A.product,cte_A.week, 
CASE WHEN cte_F.qty = cte_A.qty THEN 0 
ELSE 100.0*(cte_A.qty - cte_F.qty)/cte_F.qty As Performance 
FROM cte_A join cte_F 
ON cte_A.product=cte_F.product 
AND cte_A.week=cte_F.week