累计计数

Question

我想知道一个计数上的累积总和是否可能。我想研究的一个例子是今年影响美国的风暴。我想要一个列出 2014 年月份的结果集，以及在该月之前影响美国的风暴的累积总和。我希望得到 3 列的内容 - Month、NumberofStorms 和 CumulativeSum，如下所示：

Month        NumberofStorms     CumulativeSum
-----------------------------------------------
Jan              2                   2
Feb              1                   3
Mar              1                   4
Apr              0                   4
May              0                   4
Jun              0                   4
Jul              0                   4
Aug              0                   4    
Sep              1                   5
Oct              3                   8
Nov              5                  13 
Dec              8                  21

我怎样才能做到这一点？

Answer 1

你可以给SUM()添加一个窗口：

SELECT "Month"
      ,"NumberofStorms"
      ,SUM("NumberofStorms") OVER(ORDER BY "Month" ROWS BETWEEN unbounded preceding and current row) AS CumulativeSum
FROM Table1

演示：SQL Fiddle

注意：在演示中月份是按字母顺序排序的，因为这只是一个示例场景，所以不必费心输入真实日期。

更新：假设基表只是 Month 和 StormName，您需要一个 cte/子查询来首先获得一个计数，然后您可以在上面的运行总数中使用它：

SELECT "Month"
      ,"NumberofStorms"
      ,SUM("NumberofStorms") OVER(ORDER BY "Month" ROWS BETWEEN unbounded preceding and current row) AS CumulativeSum
FROM (SELECT "Month",COUNT("Name") AS "NumberofStorms"
      FROM Weather
      GROUP BY "Month"
      ) AS Sub

Answer 2

您可以在没有子查询的情况下进行累积和。如果NumberOfStorms 是原始数据的一部分：

SELECT "Month", "NumberofStorms",
       SUM(NumberofStorms) OVER (ORDER BY "Month" ROWS BETWEEN unbounded preceding and current row) AS CumulativeSum
FROM Weather
GROUP BY "Month", NumberOfStorms;

而且，在许多数据库中，row 子句是不必要的，因此以下内容也应该起作用：

SELECT "Month", "NumberofStorms",
       SUM(NumberofStorms) OVER (ORDER BY "Month") as CumulativeSum
FROM Weather
GROUP BY "Month", NumberOfStorms;

如果NumberOfStorms 真的来自count(*)，那么你会这样做：

SELECT "Month", count(*) as "NumberofStorms",
       SUM(count(*)) OVER (ORDER BY "Month") as CumulativeSum
FROM Weather
GROUP BY "Month";

一开始看累积和和窗口/解析函数的组合看起来有点奇怪，很快就会习惯。