Union—多表合并之行合并

Union (all)语句

select 列名    from 表A

union （all）

select 列名    from 表B

注意事项：

• 合并的表的列数必须是一致的，也就是表的列数要相同
• union 与union all 对两个查询结果合并的时候，如果结果有重复，union 会把重复的值删除；
• union 与union all不能与order by同时使用，如果要对结果进行排序，可以用子查询

案例解析

统计20170703—20170709周内每天及本周累计销售金额、订单量、会员数、订单占比

1. 统计时间段内每天的累计销售金额、订单量、会员数
2. 统计本周累计销售金额、订单量、会员数
3. 订单占比
4. union合并表
5. 对比 union all合并表——因为没有重复数据，所以两表合并结果一样

-- 1、统计时间段内每天的累计销售金额、订单量、会员数
SELECT DATE_FORMAT(dimDateID,'%W') as week_1      /*把时间转化为星期模式*/
      ,SUM(AMT) as money
      ,COUNT(DISTINCT salesID) as num_order       /*一个会员一周可以下多单，所以要统计会员数，需要去重*/
      ,COUNT(DISTINCT dimMemberID) num_member
FROM dw.fct_sales
where dimDateID between '20170703' and '20170709'
and dimMemberID <>0
group by DATE_FORMAT(dimDateID,'%W') ;
-- 2、统计本周累计销售金额、订单量、会员数
SELECT SUM(AMT) as total_money
      ,COUNT(DISTINCT salesID) as total_num_order
      ,COUNT(DISTINCT dimMemberID) total_num_member
FROM dw.fct_sales
where dimDateID between '20170703' and '20170709'
and dimMemberID <>0;
-- 3、订单占比
SELECT DATE_FORMAT(dimDateID,'%W') AS week_1
      ,CONCAT(ROUND( COUNT(DISTINCT salesID)/(SELECT COUNT(DISTINCT salesID)
                                FROM dw.fct_sales
                                where dimDateID between '20170703' and '20170709'
                                and dimMemberID <>0),4)*100,'%') as order_rate
FROM dw.fct_sales
where dimDateID between '20170703' and '20170709'
and dimMemberID <>0
group by DATE_FORMAT(dimDateID,'%W') ;
-- 4、union合并表
SELECT DATE_FORMAT(dimDateID,'%W') AS week_1 
      ,SUM(AMT) as money
      ,COUNT(DISTINCT salesID) as num_order
      ,COUNT(DISTINCT dimMemberID) num_member
      ,CONCAT(ROUND( COUNT(DISTINCT salesID)/(
       SELECT COUNT(DISTINCT salesID)
       FROM dw.fct_sales
       where dimDateID between '20170703' and '20170709'
       and dimMemberID <>0),4)*100,'%') as order_rate
FROM dw.fct_sales
where dimDateID between '20170703' and '20170709'
and dimMemberID <>0
group by DATE_FORMAT(dimDateID,'%W')
UNION 
SELECT week(dimDateID,1)               /*为了保证列数一样，week返回日期为一年中的第几周 weel(date,1):从周一开始为第一天*/
      ,SUM(AMT) as total_money
      ,COUNT(DISTINCT salesID) as total_num_order
      ,COUNT(DISTINCT dimMemberID) total_num_member
      ,'100%' as total                 /*为确保列数一样*/
FROM dw.fct_sales
where dimDateID between '20170703' and '20170709'
and dimMemberID <>0
group by week(dimDateID,1);            /*出现汇总函数需要进行分组*/
-- 5、对比 union all合并表——因为没有重复数据，所以两表合并结果一样
SELECT DATE_FORMAT(dimDateID,'%W') AS week_1 
      ,SUM(AMT) as money
      ,COUNT(DISTINCT salesID) as num_order
      ,COUNT(DISTINCT dimMemberID) num_member
      ,CONCAT(ROUND( COUNT(DISTINCT salesID)/(
       SELECT COUNT(DISTINCT salesID)
       FROM dw.fct_sales
       where dimDateID between '20170703' and '20170709'
       and dimMemberID <>0),4)*100,'%') as order_rate
FROM dw.fct_sales
where dimDateID between '20170703' and '20170709'
and dimMemberID <>0
group by DATE_FORMAT(dimDateID,'%W')
UNION ALL 
SELECT week(dimDateID,1)               /*为了保证列数一样，week返回日期为一年中的第几周 weel(date,1):从周一开始为第一天*/
      ,SUM(AMT) as total_money
      ,COUNT(DISTINCT salesID) as total_num_order
      ,COUNT(DISTINCT dimMemberID) total_num_member
      ,'100%' as total                 /*为确保列数一样*/
FROM dw.fct_sales
where dimDateID between '20170703' and '20170709'
and dimMemberID <>0
group by week(dimDateID,1);            /*出现汇总函数需要进行分组*/

• 区分union 和 union all ，利用重复数据对比，合并两个一模一样的表 

-- union
SELECT DATE_FORMAT(dimDateID,'%W') as week_1      /*把时间转化为星期模式*/
      ,SUM(AMT) as money
      ,COUNT(DISTINCT salesID) as num_order       /*一个会员一周可以下多单，所以要统计会员数，需要去重*/
      ,COUNT(DISTINCT dimMemberID) num_member
FROM dw.fct_sales
where dimDateID between '20170703' and '20170709'
and dimMemberID <>0
group by DATE_FORMAT(dimDateID,'%W')
UNION 
SELECT DATE_FORMAT(dimDateID,'%W') as week_1      /*把时间转化为星期模式*/
      ,SUM(AMT) as money
      ,COUNT(DISTINCT salesID) as num_order       /*一个会员一周可以下多单，所以要统计会员数，需要去重*/
      ,COUNT(DISTINCT dimMemberID) num_member
FROM dw.fct_sales
where dimDateID between '20170703' and '20170709'
and dimMemberID <>0
group by DATE_FORMAT(dimDateID,'%W') ;
-- union all
SELECT DATE_FORMAT(dimDateID,'%W') as week_1      /*把时间转化为星期模式*/
      ,SUM(AMT) as money
      ,COUNT(DISTINCT salesID) as num_order       /*一个会员一周可以下多单，所以要统计会员数，需要去重*/
      ,COUNT(DISTINCT dimMemberID) num_member
FROM dw.fct_sales
where dimDateID between '20170703' and '20170709'
and dimMemberID <>0
group by DATE_FORMAT(dimDateID,'%W')
UNION ALL 
SELECT DATE_FORMAT(dimDateID,'%W') as week_1      /*把时间转化为星期模式*/
      ,SUM(AMT) as money
      ,COUNT(DISTINCT salesID) as num_order       /*一个会员一周可以下多单，所以要统计会员数，需要去重*/
      ,COUNT(DISTINCT dimMemberID) num_member
FROM dw.fct_sales
where dimDateID between '20170703' and '20170709'
and dimMemberID <>0
group by DATE_FORMAT(dimDateID,'%W') ;

• 多表合并中的排序问题 order by——把合并后的表作为一个临时表，再进行排序

注意临时表需要命名

排序要对新的列名进行排序

SELECT sn.*
FROM (
      SELECT DATE_FORMAT(dimDateID,'%W') as week_1      /*把时间转化为星期模式*/
            ,SUM(AMT) as money 
            ,COUNT(DISTINCT salesID) as num_order       /*一个会员一周可以下多单，所以要统计会员数，需要去重*/
            ,COUNT(DISTINCT dimMemberID) num_member
      FROM dw.fct_sales
      where dimDateID between '20170703' and '20170709'
      and dimMemberID <>0
      group by DATE_FORMAT(dimDateID,'%W')
      UNION 
      SELECT DATE_FORMAT(dimDateID,'%W') as week_1      /*把时间转化为星期模式*/
            ,SUM(AMT) as money 
            ,COUNT(DISTINCT salesID) as num_order       /*一个会员一周可以下多单，所以要统计会员数，需要去重*/
            ,COUNT(DISTINCT dimMemberID) num_member
      FROM dw.fct_sales
      where dimDateID between '20170703' and '20170709'
      and dimMemberID <>0
      group by DATE_FORMAT(dimDateID,'%W') ) as sn
order by money desc;                                    /*注意要对money排序，而不是SUM(AMT)*/
-- order by SUM(AMT) desc;                              /*会报错，需要用临时表的列名，并且要对临时表取列名，否则也会报错*/