Question

使用SQL Server，我需要为每行的值获取一个包含-1,0,1,2的列计数。每个表有不同数量的列（一个表有55个字段）

-1 value = no answer or unknown (unknown_count)
0  value = ok (ok_count)
1  value = bad  (bad_count)
2  value = not applicable  (na_count)

第1行看起来像这样

rowid        name       field_1       field_2     field_3  field_4
1            line_1     -1            1           2        1
2            line_2     2             1           -1       0

等...

我希望看到结果

rowid     na_count     ok_count     bad_count     unknown_count
1         1            0            2             1
2         1            1            1             1

额外的信用问题....我需要计算用于构建计数的所有字段（因此我可以提供百分位报告）

Answer 1

您需要UNPIVOT和一些聚合的组合。将数据展开为更合理的形式，然后按rowid分组，总结各种值：

with data AS
(
  select 
    rowid,
    field,
    value
  from
   ( SELECT rowid, field_1,field_2,field_3,field_4
     FROM MyTable) p
  UNPIVOT
    ( value FOR field IN (field_1,field_2,field_3,field_4) ) as unpvt
)
SELECT 
  rowid,
  SUM(CASE WHEN value = 2 THEN 1 ELSE 0 END) AS na_count,
  SUM(CASE WHEN value = 0 THEN 1 ELSE 0 END) AS ok_count,
  SUM(CASE WHEN value = 1 THEN 1 ELSE 0 END) AS bad_count,
  SUM(CASE WHEN value = -1 THEN 1 ELSE 0 END) AS unknown_count
from data
group by rowId

实例：http://www.sqlfiddle.com/#!6/b702c/1

Answer 2

以Jamiec的答案和我之前使用的一些代码为基础;如果您不想在查询中键入所有列名，可以使用动态SQL来构建查询，如下所示：

declare @tab nvarchar(max)
set @tab = N'your_table' -- change to your table name

declare @cols nvarchar(max) 
select @cols = coalesce(@cols+N',', N'') + quotename(c.name) from syscolumns c
inner join sysobjects o on c.id = o.id and o.xtype = 'u'
where o.name = @tab 
and c.name not in ('rowid', 'name') -- exclude the columns that don't hold data values
order by c.colid

declare @sql nvarchar(max)  
select @sql = N'
select 
    rowid, 
    sum(case when val = 2 then 1 else 0 end) as ''na_count'',
    sum(case when val = 0 then 1 else 0 end) as ''ok_count'',
    sum(case when val = 1 then 1 else 0 end) as ''bad_count'',
    sum(case when val = -1 then 1 else 0 end) as ''unknown_count'',
    count(*) as column_count
from (select rowid, ' + @cols + N' from ' + @tab + N') as src
unpivot (val for col in (' + @cols + N')) as unpvt
group by rowid'

exec sp_executesql @sql

Answer 3

我认为在动态脚本中总结所有列会更有效，而不是使用unpivot。在一百万行中，相对于执行计划中的批处理，我获得了16％和84％（此代码的输出@query与unpivot代码相比）。

您可以使用相同的逻辑来获取每列的百分比。如果您希望我也为您提供代码，请与我们联系。

--DROP TABLE TMP_Test
CREATE TABLE TMP_Test
(
    rowid INT PRIMARY KEY IDENTITY(1,1) 
    , name varchar(10)
    , field_1 INT
    , field_2 INT
    , field_3 INT
    , field_4 INT
)

INSERT INTO TMP_Test
SELECT name = 'line_1', field_1 = -1, field_2=1, field_3=2, field_4=1 
UNION ALL
SELECT name = 'line_2', field_1 = 2, field_2=1, field_3=-1, field_4= 0

/*
WHILE((SELECT COUNT(*) FROM TMP_Test) < 1000000)
BEGIN
    INSERT INTO TMP_Test
    SELECT name, field_1, field_2, field_3, field_4  FROM TMP_Test
END
*/


GO

DECLARE @query VARCHAR(MAX) = '';
DECLARE @schema VARCHAR(128) = 'dbo';
DECLARE @table VARCHAR(128) = 'TMP_Test';

DECLARE @na_count VARCHAR(max) = '';
DECLARE @ok_count VARCHAR(max) = '';
DECLARE @bad_count VARCHAR(max) = '';
DECLARE @unknown_count VARCHAR(max) = '';

SELECT 
@na_count = @na_count + IIF(@na_count = '', '', ' + ') + 'IIF(' + COLUMN_NAME + ' = 2, 1, 0)'
, @ok_count = @ok_count + IIF(@ok_count = '', '', ' + ') + 'IIF(' + COLUMN_NAME + ' = 0, 1, 0)'
, @bad_count = @bad_count + IIF(@bad_count = '', '', ' + ') + 'IIF(' + COLUMN_NAME + ' = 1, 1, 0)'
, @unknown_count = @unknown_count + IIF(@unknown_count = '', '', ' + ') + 'IIF(' + COLUMN_NAME + ' = -1, 1, 0)'
FROM INFORMATION_SCHEMA.COLUMNS WHERE 
TABLE_SCHEMA = @schema
AND TABLE_NAME = @table
AND COLUMN_NAME NOT IN ('rowid', 'name')
ORDER BY ORDINAL_POSITION;

SET @query = '
SELECT
rowid
, na_count = ' + @na_count + '
, ok_count = ' + @ok_count + '
, bad_count = ' + @bad_count + '
, unknown_count = ' + @unknown_count + '
FROM [' + @schema + '].[' + @table + ']';

PRINT(@query);
EXEC(@query);

获取具有单个行的-1,0,1,2值的列的计数

3 个答案: