SQL Server计算最常见数据的实例

Question

我不相信我需要做的事情太复杂了，但是嘿，这是星期一早上我可以使用一些建议。基本上我在SQL Server中以下列格式进行聚合查询：

SELECT [Data1], COUNT(*), MAX([Data2])
FROM [Source]
GROUP BY [Data1]

但是我也需要第4场。一个字段，计算值MAX([Data2])出现的次数。

我将在这里做一个简单的例子：

|Data 1|Data 2|
|1     |x     |
|3     |p     |
|1     |z     |
|3     |f     |
|1     |x     |
|1     |b     |
|2     |h     |
|1     |o     |
|2     |h     |
|1     |x     |
|3     |f     |
|2     |h     |
|1     |z     |

需要产生输出：

|Data1|Count|Max|Occurances|
|1    |7    |x  |3         |
|2    |3    |h  |3         |
|3    |3    |f  |2         |

任何帮助，将不胜感激。

Answer 1

您正在寻找的是统计中的名称。您需要模式的模式和频率。

我会使用两级聚合来处理这个问题：

select data1, sum(cnt) as cnt,
       max(case when seqnum = 1 then data2 end) as mode,
       max(cnt) as mode_cnt
from (select data1, data2, count(*) as cnt,
             row_number() over (partition by data1 order by count(*) desc) as seqnum
      from t
      group by data1, data2
     ) t
group by data1;

Answer 2

这样的事情应该有所帮助 -

;WITH CTE AS
(SELECT [Data1], COUNT(*) AS Data_Count, MAX([Data2]) AS Data_Max
FROM [Source]
GROUP BY [Data1])
SELECT [Data1],Data_Count,Data_Max,COUNT(Data_Max) AS Occurances
FROM CTE 
GROUP BY [Data1],Data_Count,Data_Max

注意：我会避免使用SQL Keywords作为列名/别名。

Answer 3

您可以尝试使用cteto的窗口功能。

使用CTE通过COUNT和Occurances专栏获得[Data 1] [Data 2]。然后使用另一个CTE2通过Occurances desc生成行号，这意味着最大金额。

;with cte as (
SELECT [Data 1],
       [Data 2],
       COUNT(*) OVER (PARTITION BY [Data 1] ORDER BY [Data 1]) [count],
       COUNT(*) OVER (PARTITION BY [Data 2],[Data 1] ORDER BY [Data 2]) Occurances
FROM T
),cte2 as (
    SELECT  
        [Data 1], 
        [Data 2],
        [count],
        Occurances,row_number() over(partition by [Data 1] order by Occurances desc) rn
    FROM cte
)

select  [Data 1], 
        [Data 2],
        [count], 
        Occurances
from cte2
where rn = 1

sqlfiddle

结果

Data 1  Data 2  count   Occurances
1       x         7     3
2       h         3     3
3       f         3     2

Answer 4

你可以试试这个。

DECLARE @Source AS TABLE([Data1] INT,[Data2] VARCHAR(5))
INSERT INTO @Source VALUES
(1, 'x'), 
(3, 'p'),
(1, 'z'),
(3, 'f'),
(1, 'x'),
(1, 'b'),
(2, 'h'),
(1, 'o'),
(2, 'h'),
(1, 'x'),
(3, 'f'),
(2, 'h'),
(1, 'z')

;WITH T AS (
SELECT [Data1] , 
    COUNT(*) OVER(PARTITION BY [Data1]) [Count], 
    [Data2],    
    COUNT(*) OVER(PARTITION BY [Data2]) Occurances
FROM @Source
)
SELECT TOP 1 WITH TIES [Data1], [Count], [Data2] [Max], Occurances
FROM T 
ORDER BY ROW_NUMBER() OVER (PARTITION BY [Data1] ORDER BY Occurances DESC) 

结果：

Data1       Count       Max   Occurances
----------- ----------- ----- -----------
1           7           x     3
2           3           h     3
3           3           f     2