SELECT column1, SUM(Avg_col3) AS total
FROM (
    SELECT column1, column2, AVG(column3) AS avg_col3
    FROM t
    GROUP BY column1, column2
) A
GROUP BY column1;

您可以通过两个聚合来实现这一点：

SELECT Column1, SUM(Column3Avg) AS Column3
FROM
(
    SELECT Column1, Column2, AVG(Column3) AS Column3Avg
    FROM yourTable
    GROUP BY Column1, Column2
) t
GROUP BY
    Column1;

此方法首先通过第一列和第二列进行聚合，以生成如下所示的中间结果：

Column1 | Column2 | Column3Avg
H1      | A1      | 10
H1      | A2      | 30

然后，我们单独用第一列对上述数据进行聚合，并对第3列的平均值求和。

如果有这样重复的数字，你的数据模型看起来真的一团糟。我建议：

select column1, sum(column3)
from (select distinct column1, column2, column3
      from t
     ) t
group by column1;

为了 SQL Server 使用以下查询获得所需的结果。

WITH CTE AS (
    SELECT  Column1,
            AVG(Column3) AS Column3
    FROM Table1
    GROUP BY Column1, Column2
)
SELECT Column1, SUM(Column3) AS Column3
FROM CTE
GROUP BY Column1

请运行以下代码。你会得到你的结果的。

Select *
into #tmp
from (
    select 'h1' as Column1, 'A1' as Column2, 10 as Column3
    union all
    select 'h1' as Column1, 'A1' as Column2, 10 as Column3
    union all
    select 'h1' as Column1, 'A2' as Column2, 30 as Column3
    union all
    select 'h1' as Column1, 'A2' as Column2, 30 as Column3
    union all
    select 'h1' as Column1, 'A2' as Column2, 30 as Column3
)r
select * from #tmp
Select r.Column1, sum(r.Column3) Column3
from (
    select Column1, Column2, avg(Column3) Column3
    from #tmp
    group by Column1, Column2
 )r
 group by r.Column1