分组，多个计数，多个表SQL Server

Question

I have two tables as following:

First table: 2016

Os          |   Count
------------+----------
Windows 7         8
Windows 7         9
Windows 7        20
Windows 8        30
Linux            15

Second table: 2017

Os          |   Count
------------+----------
Windows 7        35
Windows 7        11
Windows 8        10
Windows 8         8
Linux            10
Ubuntu            3

I try by myself but my challenge is to use GROUP BY function in meanwhile I'm counting two fields from two different tables, I got always error.

Thank you in advance,

Update:

I'm sorry I did a mistake in explaining my request: what I need is to write a query to get a result like this:

    --OS--         --2016--    --2017--
    ------------------------------------
    Windows 7        37             46
    Windows 8        30             18
    Linux            15             10
    Ubuntu           0              3

Answer 1

You can do simple group by as below:

select a.OS, sum(count) from (
    select * from your2016table
        union all 
    select * from your2017table
    ) a
    group by a.OS

For that you can use year as group by and pivot

;with cte as (
select a.OS, a.[year], SumCt = sum(count) from (
    select *, 2016 as [year] from your2016table
        union all 
    select *, 2017 as [year] from your2017table
    ) a
    group by a.OS, a.[year]
)
select * from cte 
pivot (max(sumct) for [year] in ([2016], [2017])) p

Answer 2

Without a UNION ALL you can do this:

SELECT
    COALESCE( [2016].[OS], [2017].[OS] ) AS [OS],
    ( [2016].[Count] + [2017].[Count] ) AS [Count]
FROM
    [2016]
    FULL OUTER JOIN [2017] ON [2016].[OS] = [2017].[OS]

This also eliminates the need for GROUP BY too, assuming the OS columns contain no duplicates.

Duplicates would need to be eliminated through GROUP BY (rather than SELECT DISTINCT ), but the same structure applies:

SELECT
    COALESCE( [2016-D].[OS], [2017].[OS] ) AS [OS],
    ( [2016-D].[Count] + [2017].[Count] ) AS [Count]
FROM
    (
        SELECT [OS], SUM( [Count] ) FROM [2016] GROUP BY [OS]
    ) AS [2016-D]
    FULL OUTER JOIN
    (
        SELECT [OS], SUM( [Count] ) FROM [2017] GROUP BY [OS]
    ) AS [2017-D]
        ON [2016].[OS] = [2017].[OS]

...but then it gets quite unwieldy, and the UNION ALL approach becomes simpler!

Update:

The OP has amended their question to say they want separate columns - so this is where using an OUTER JOIN becomes the ideal solution:

SELECT
    COALESCE( [2016-D].[OS], [2017-D].[OS] ) AS [OS],
    [2016-D].[Count] AS [2016],
    [2017-D].[Count] AS [2017]
FROM
    (
        SELECT
            [OS],
            SUM( [Count] ) AS [Count]
        FROM
            [2016]
        GROUP BY
            [OS]
    ) AS [2016-D]
    FULL OUTER JOIN
    (
        SELECT
            [OS],
            SUM( [Count] ) AS [Count]
        FROM
            [2017]
        GROUP BY
            [OS]
    ) AS [2017-D]
        ON [2016].[OS] = [2017].[OS]
ORDER BY
    [OS]