SQL Server - 派生列使用第二行中自己的先前值

Question

我需要在 SQL Server 中创建一个派生列 ColumnC。规则是 ColumnC 的第一行是 ColumnA 和 ColumnB 的乘积。以下行是ColumnC中上一行的乘积，按年份乘以ColumnA * ColumnC。

+-----------+--------------------------+------------------------+
| **Year**  |        **ColumnA**       |      **ColumnB**       |
+-----------+--------------------------+------------------------+
|  2020     |        0.987441          |        0.001039        |
+-----------+--------------------------+------------------------+
|  2021     |        0.975952          |        0.001117        |
+-----------+--------------------------+------------------------+
|  2022     |        0.965471          |        0.001206        |
+-----------+--------------------------+------------------------+
|  2023     |        0.955950          |        0.001293        |
+-----------+--------------------------+------------------------+
|  2024     |        0.947347          |        0.001387        |
+-----------+--------------------------+------------------------+
|  2025     |        0.939604          |        0.001488        |
+-----------+--------------------------+------------------------+
|  2026     |        0.933461          |        0.001596        |
+-----------+--------------------------+------------------------+
|  2027     |        0.922700          |        0.001710        |
+-----------+--------------------------+------------------------+
|  2028     |        0.914439          |        0.001959        |
+-----------+--------------------------+------------------------+
|  2029     |        0.900277          |        0.002134        |
+-----------+--------------------------+------------------------+

为了保持示例简单，我简化了计算和表格，因此，我需要按照描述计算列。

遵守以下算法对我来说很重要：ColumnC 的第一行是 ColumnA abd ColumnB 的乘积，接下来的行，从第二行开始，使用 ColumnC 的前一行，按 Year × ColumnB × ColumnC 排序。

让我陷入困境的最大问题是我不知道如何获取 ColumnC 的先前值。我不能使用LAG，因为该列还不存在。

SELECT
    ColumnA,
    ColumnB,
    /* First row okay. But the following rows should 
     use the previous value of ColumnC times ColumnA times ColumnB
     that is ColumnC(Year-1) * ColumnA * ColumnB
    CASE WHEN ROW_NUMBER() OVER (ORDER BY Year) = 1 THEN
        ColumnA * ColumnB
    ELSE
        LAG(ColumnC) OVER (ORDER BY Year) * ColumnA * ColumnB
    END AS ColumnC
    */
    (ColumnA * ColumnB) AS ColumnC
FROM TableA

这行不通：

CASE WHEN ROW_NUMBER() OVER (ORDER BY Year ASC) = 1 THEN
    ColumnA * ColumnB
ELSE
    LAG(ColumnC) OVER (ORDER BY Year ASC) * ColumnA * ColumnB
END AS ColumnC

请帮助我理解我的问题。

Useless SQL Fiddle

从ColumnC计算第一行：

从 ColumnC 的第二行开始计算：

结果：

Answer 1

数字变得非常小非常快，但这符合您的描述：

select a.*, 
       columna * columnb,
       exp(sum(log(columna * columnb)) over (order by year))
from tableA a

Answer 2

根据提供的信息，这是您要查找的内容吗？

DROP TABLE IF EXISTS #TableA

CREATE TABLE #TableA (
    [Year] INT,
    ColumnA DECIMAL(8, 6),
    ColumnB DECIMAL(8, 6)
)

INSERT INTO #TableA (
  [Year],
  ColumnA,
  ColumnB
)
VALUES
    (2020, 0.987441, 0.001039),
    (2021, 0.975952, 0.001117),
    (2022, 0.965471, 0.001206),
    (2023, 0.955950, 0.001293),
    (2024, 0.947347, 0.001387),
    (2025, 0.939604, 0.001488),
    (2026, 0.933461, 0.001596),
    (2027, 0.922700, 0.001710),
    (2028, 0.914439, 0.001959),
    (2029, 0.900277, 0.002134);

    SELECT *, CASE WHEN ROW_NUMBER() OVER (ORDER BY Year ASC) = 1 THEN ColumnA * ColumnB
              ELSE LAG(ColumnB) OVER (ORDER BY Year ASC) * ColumnA * ColumnB
               END AS ColumnC
      FROM #TableA