SQL 获取季度值

Question

我想从包含 Q3 和 Q4 数据的数据集中获取 QoQ，该数据集也有一个报告日期列，每行应该有每个会计月的 QoQ 值（由报告日期表示），Q4 应该比较与 Q3 相比，但我的陈述似乎只是在同一季度内进行比较，即 Q4 与 Q4 进行比较，而不是 Q4 与 Q3 进行比较 ..

我正在使用 lag 函数，但不确定我做错了什么，如果有人可以请看下面的代码。

 SELECT [Year], 
  [SalesDate] as Report_Date,
       [Quarter], 
       Sales,
     
       LAG(Sales,  1, 0) OVER(
       PARTITION BY [Year] ,[Quarter]
       ORDER BY [Year], 
                [Quarter],
                salesDate
                ASC) AS [QuarterSales_Offset],
                sales - LAG(Sales) OVER(
       PARTITION BY [Year] ,[Quarter]
       ORDER BY [Year], 
                [Quarter],
                salesDate
                ASC) as diff,
Case When 
LAG(Sales,1,0) OVER(
       PARTITION BY [Year],[Quarter]
       ORDER BY [Year], 
                [Quarter],
                salesDate
                ASC) = 0 then null else

(
sales - LAG(Sales,1,0) OVER(
       PARTITION BY [Year],[Quarter]
       ORDER BY [Year], 
                [Quarter],
                salesDate
                ASC))/ LAG(Sales,1,0) OVER(
       PARTITION BY [Year],[Quarter]
       ORDER BY [Year], 
                [Quarter],
                salesDate
                ASC) end as QoQ
FROM dbo.ProductSales_2;

查询输出：

Answer 1

由于 LAG() 在 1 偏移量 returns 前一行并且您的数据处于月级别，因此您实际上比较了每个季度的月度数据。考虑一种不同的方法，例如按季度和按月连接数据的两个子集。

QuarterMonth 列可以用 ROW_NUMBER() 表达式计算（即每个季度内的运行月数）。由于可能会出现销售数据中的月份差异，因此请使用与您的财政年度对齐的 year_quarter_month 日历 table。总而言之，这允许按列比较 FY Q4 的第一个月份 (2020-08-31) 和 FY Q3 的第一个月份 (2020-05-31)。

WITH unit AS (
   SELECT yqm.[Year]
        , yqm.[Quarter]
        , yqm.[Month]
        , COALESCE(p.[Report_Date], DATEADD(DAY, -1, DATEFROMPARTS(yqm.[Year], yqm.[Month]+1, 1))) AS [Report_Date]
        , p.[Sales]
   FROM year_quarter_month_table yqm
   LEFT JOIN dbo.ProductSales_2 p
      ON yq.[Year] = p.[Year]
      AND yq.[Quarter] = p.[Quarter]
      AND yq.[Month] = p.[Month]

), sub AS (
   SELECT [Year]
        , [Quarter]
        , [Month]
        , ROW_NUMBER() OVER(PARTITION BY [Year], [Quarter] 
                            ORDER BY [Report_Date]) AS [QuarterMonth]
        , [Report_Date]
        , [Sales]
   FROM unit
 )

SELECT q4.[Year]
     , q4.[Report_Date] AS Q4_Date
     , q4.[Sales] AS Q4_Sales
     , q3.[Report_Date] AS Q3_Date
     , q3.[Sales] AS Q3_Sales
     , q4.[Sales] - q3.[Sales] AS Diff
     , COALESCE((q4.[Sales] - q3.[Sales]) / q3.[Sales], 0) AS QoQ
FROM sub q4
LEFT JOIN sub q3
   ON  q4.[Year] = q3.[Year]
   AND q4.[Quarter] = 4
   AND q3.[Quarter] = 3
   AND q4.[QuarterMonth] = q3.[QuarterMonth]

您可以概括为任何 quarter-over-quarter 计算，而不仅仅是 Q3 和 Q4：

WITH sub AS (
  -- SAME CTEs AS ABOVE
)

SELECT curr_qtr.[Year]
     , curr_qtr.[Report_Date] AS Curr_Qtr_Date
     , curr_qtr.[Sales] AS Curr_Qtr_Sales
     , last_qtr.[Report_Date] AS Last_Qtr_Date
     , last_qtr.[Sales] AS Last_Qtr_Sales
     , curr_qtr.[Sales] - last_qtr.[Sales] AS Diff
     , COALESCE((curr_qtr.[Sales] - last_qtr.[Sales]) / last_qtr.[Sales], 0) AS QoQ
FROM sub curr_qtr
LEFT JOIN sub last_qtr
   ON  curr_qtr.[Year] = last_qtr.[Year]  -- ASSUMING FISCAL YEAR AND NOT CALENDAR YEAR
   AND curr_qtr.[Quarter] = last_qtr.[Quarter] + 1
   AND curr_qtr.[QuarterMonth] = last_qtr.[QuarterMonth]

SQL 获取季度值

SQL Get Quarter over Quarter values

sql

fiscal

quarter