如何合并在TPH与空值的行，并计算相关领域

Question

如果我有一个反规范化表是这样的：

FinanceList (NetSales,Expenses,Receivables,...etc,Discriminator) 

而且我还有一个表（PeriodType），我归类根据一年两种方法：

• Quarter（4周期）

  1- Period1 1月至三月

  Period2 2-从4至6月

  3- Period3从7至9月

  4- Period4从10至12月

• Yearly（1周期）从1至12月

---

示例数据：

Year per TypeId NetSales Expenses Receivables CompanyId 
2016 2 2 164547.0000 NULL  NULL  3001 
2016 2 2 NULL   NULL  50601.0000 3001 
2016 2 2 NULL   550.4110 NULL  3001 
2016 3 2 222764.0000 NULL  NULL  3001 
2015 3 2 264843.0000 NULL  NULL  3001 
2015 2 2 NULL   NULL  42049.0000 3001 
2015 1 3 NULL   NULL  32431.0000 3001 
2015 2 2 NULL   614.6200 NULL  3001 
2015 2 2 187112.0000 NULL  NULL  3001 
2014 1 3 NULL   NULL  28033.0000 3001 
2016 3 2 502757.0000 NULL  NULL  3002 
2016 3 2 NULL   NULL  56407.0000 3002 
2016 2 2 429821.0000 NULL  NULL  3002 
2016 2 2 NULL   516.0000 NULL  3002 
2016 2 2 NULL   NULL  70724.0000 3002 
2015 2 2 NULL   6092.0000 NULL  3002 
2015 2 2 NULL   NULL  96377.0000 3002 
2015 2 2 598416.0000 NULL  NULL  3002 
2015 3 2 677026.0000 NULL  NULL  3002 
2015 3 2 NULL   NULL  NULL  3002 
2015 1 3 NULL   NULL  92406.0000 3002 
2014 1 3 NULL   NULL  84243.0000 3002 

现在我面临着两个问题：

• 因为我跟着TPH（许多空在那里），所以我想合并所有 行（year,period,periodTypeId）为每个公司在一行 。

EX：代替三排，每一个台（类型）：

2016 2 2 164547.0000 NULL  NULL  3001 
    2016 2 2 NULL   NULL  50601.0000 3001 
    2016 2 2 NULL   550.4110 NULL  3001 

    2015 2 2 NULL   NULL  42049.0000 3001 
    2015 2 2 NULL   614.6200 NULL  3001 
    2015 2 2 187112.0000 NULL  NULL  3001 

我想一排这样的：

2016 2 2 164547.0000 550.4110 50601.0000 3001 
    2015 2 2 187112.0000 614.6200 42049.0000 3001 

• 如果我想获得一个等式计算如下

(The accumulated number of days in the quarter（90 or 180 0r 270 or 360）/（NetSales /（（Receivables in the specific quarter + Receivables in the previous yearly year）/ 2））

因此，例如，如果要calc下在period2 of quarter这个方程（3001）公司它将是这样的：

180/(164547.0000/((50601.0000 +32431.0000)/2)) =45 (for year 2016) 
180/(187112.0000/((42049.0000 +28033.0000)/2)) =34 (for year 2015) 

---

我的查询：

SELECT a.[Year],d.period,d.PeriodTypeId, a.NetSales,a.Expenses,a.Receivables, 
c.CompanyId 
FROM FinanceList a INNER JOIN Company c 
ON a.CompanyId = c.CompanyId 
INNER JOIN ListPeriod d 
ON d.FinanceListId = a.FinanceListId 
WHERE a.[Year] IN (2016,2015) AND d.PeriodTypeId IN(2,3) --The User Enter only Two Years ,PeriodTypeId ==>2 means quarter ,3 means yearly 
ORDER BY c.CompanyId, a.[Year] DESC 

---

我想最终结果是这样两个记录每个公司：

Year per TypeId NetSales  Expenses Receivables CompanyId equation 
2016 2 2  164547.0000 550.4110 50601.0000 3001  45 
2015 2 2  187112.0000 614.6200 42049.0000 3001  34 

Answer 1

我认为这是可以通过聚合函数来实现。使用公共表格表达式来选择搜索的时间段和年份及其前几年。处理结果和计算的返回数据。 注意：我只有以下查询语法测试。

WITH data AS (
    SELECT a.[Year], 
     d.period, 
     d.PeriodTypeId, 
     SUM(a.NetSales) AS NetSales, 
     SUM(a.Expenses) AS Expenses, 
     SUM(a.Receivables) AS Receivables, 
     c.CompanyId 
    FROM FinanceList a 
     INNER JOIN Company c 
       ON a.CompanyId = c.CompanyId 
     INNER JOIN ListPeriod d 
       ON d.FinanceListId = a.FinanceListId 
    WHERE a.[Year] IN (2016, 2015) AND d.PeriodTypeId IN(2, 3) 
     OR a.[Year] IN (2016 - 1, 2015 - 1) AND d.PeriodTypeId = 3 -- previous years 
    GROUP BY a.[Year], d.period, d.PeriodTypeId, c.CompanyId 
) 
SELECT [Year], 
     period, 
     PeriodTypeId, 
     NetSales, 
     Expenses, 
     Receivables, 
     CompanyId, 
     CASE 
     WHEN PeriodTypeId = 2 
     THEN (period * 90)/(NetSales/((Receivables + (SELECT Receivables 
                  FROM data T2 
                  WHERE T2.[Year]  = T1.[Year] - 1 
                   AND T2.period  = 1 
                   AND T2.PeriodTypeId = 3 
                   AND T2.CompanyId = T1.CompanyId) /* Receivables previous year */)/2)) 
     ELSE NULL 
     END AS equation 
    FROM data T1 
WHERE [Year] IN (2016, 2015) AND PeriodTypeId IN(2, 3) 
ORDER BY CompanyId, [Year] DESC