Question

我正在尝试压缩数据库表中的数据，该表包含具有不同列数据的唯一记录的多个实例。

我想为每个列选择最高出现值特殊的独特记录

但是我的SQL事务无法正常工作。

[dataBase1].[dbo].[table1]有数十万条记录，其中有几列(Name, Place, etc.)。

[dataBase1].[dbo].[table2]包含[table1]中唯一名称的列表以及其余为空的列（页头等）的标题。

我尝试了以下代码。

DECLARE @name varchar(max);
DECLARE @place varchar(max);

DECLARE db_cursor SCROLL CURSOR FOR 
     SELECT [Name] 
     FROM [dataBase1].[dbo].[table2];

OPEN HostName_cursor

FETCH NEXT FROM db_cursor INTO @name;

WHILE @@FETCH_STATUS = 0
BEGIN
     SELECT DISTINCT TOP(1) @place = [Place] 
     FROM [dataBase1].[dbo].[table1] 
     WHERE [Name] = @name 
       AND [Place] IS NOT NULL AND [Place] <> '' 
       AND (EXISTS  (SELECT [Place], COUNT (*) AS TOTAL 
                     FROM [dataBase1].[dbo].[table1] 
                     GROUP BY [Place])) 
     GROUP BY [Place];

     UPDATE [dataBase1].[dbo].[table2] 
     SET [Place] = @place 
     WHERE [Name] = @name;      

     SET @place = '';

     FETCH NEXT FROM db_cursor INTO @name
END

特定唯一[Place]的

[Name]列具有 53 值，最高重复值计数为 3 。本质上，我想为每个唯一的[Name]自动执行以下SQL事务。

SELECT DISTINCT TOP 1 
    [Place], COUNT (*) TOTAL 
FROM 
    [dataBase1].[dbo].[table1] 
WHERE 
    [Name] = 'xxxxxx' 
    AND [Place] IS NOT NULL AND [Place] <> '' 
GROUP BY [Place] 
ORDER BY TOTAL DESC;

Answer 1

这可以通过许多步骤来完成，每个步骤都在接下来的步骤中进行。您想一次处理所有名称和地点。

首先，您要统计每个名称的数量，地点组合，因此请按名称和地点分组并计算地点。您的查询将如下所示

SELECT name, place, COUNT(place) as placecount
FROM table1
GROUP BY name, place

现在，您需要找到计数最高的一个，如果是平局，则需要按字母顺序查找第一个。您可以通过对上述结果进行ROW_NUMBER，重新启动名称的计数（分区），然后按地点计数，然后按地点排序来解决关系，来完成此操作。使用CTE（您也可以将其作为子查询来执行），看起来就像

WITH places as (
  SELECT name, place, COUNT(place) as placecount
  FROM table1
  GROUP BY name, place
)
SELECT name, place, ROW_NUMBER() OVER (PARTITION BY name ORDER BY placecount, place) as RN
FROM places

如果查看这些数据，则任何给定名称的位置都应该在RN为1的行上。这样，您就可以通过这样的查询获得所需的最终数据

WITH places as (
  SELECT name, place, COUNT(place) as placecount
  FROM table1
  GROUP BY name, place
), orderplaces as (
  SELECT name, place, ROW_NUMBER() OVER (PARTITION BY name ORDER BY placecount, place) as RN
  FROM places
)
Select name, place
FROM orderplaces
WHERE RN = 1

由于要使用此位置数据更新table2而不是查看它，您将在最终查询中加入table2并进行更新，类似这样

WITH places as (
  SELECT name, place, COUNT(place) as placecount
  FROM table1
  GROUP BY name, place
), orderplaces as (
  SELECT name, place, ROW_NUMBER() OVER (PARTITION BY name ORDER BY placecount, place) as RN
  FROM places
)
UPDATE T2 set place = OP.place
FROM orderplaces OP
   INNER JOIN table2 T2 on T2.name = OP.name
WHERE RN = 1;

需要有关SQL事务的协助以压缩数据库中的数据

1 个答案: