Oracle 行列转换函数pivot、unpivot的使用(二)

一、行转列pivot

关键函数pivot,其用法如下 pivot(聚合函数 for 列名 in(类型))

select * from table_name pivot(max(column_name)                            --行转列后的列的值value,聚合函数是必须要有的
                               for column_name in(value_1,value_2,value_3)     --需要行转列的列及其对应列的属性1/2/3
                                     )

1、首先举一个简单的例子,创建一个数据表

1 create table tmp as select * from (
2 select '张三' student,'语文' course ,78 score from dual union all 
3 select '张三','数学',87 from dual union all 
4 select '张三','英语',82 from dual union all 
5 select '张三','物理',90 from dual union all 
6 select '李四','语文',65 from dual union all 
7 select '李四','数学',77 from dual union all 
8 select '李四','英语',65 from dual union all 
9 select '李四','物理',85 from dual);

先使用decode或case when方法

 1 select 
 2   student, 
 3   max(decode(course, '语文', score)) 语文, 
 4   max(decode(course, '数学', score)) 数学, 
 5   max(decode(course, '英语', score)) 英语, 
 6   max(decode(course, '物理', score)) 物理, 
 7   sum(score) total 
 8 from tmp
 9 group by student;
10 -----------------------------------------
11 select 
12   student,
13   max(case when course = '语文' then score end) 语文, 
14   max(case when course = '数学' then score end) 数学, 
15   max(case when course = '英语' then score end) 英语, 
16   max(case when course = '物理' then score end) 物理, 
17   sum(score) total 
18 from tmp
19   group by student;

pivot的使用

1 select t.*,  
2   (t.语+t.数+t.外+t.物) as total  
3 from  
4   (select *  
5   from tmp pivot ( max(score) for course in ('语文' as 语 , '数学' as 数, '英语' as 外,'物理' as 物) )  
6   ) t;

结果同上

2、实际开发遇到的问题

有一张目标值表,年、月、日的值都是分开多行显示,现需合并成一行显示,具体数据如下:(type:1-->日,2-->月,3-->年;targetvalue:目标值)

select * from MOVEBI.T_GMS_MBI_TARGET_DATA where targetcode = '31227061'

 

此数据必须先进性处理,要保证数据可以聚合成一条,若直接使用会出现下列情况:

select * from MOVEBI.T_GMS_MBI_TARGET_DATA pivot(max(targetvalue) for type in (1 day_value,2 mon_value,3 year_value)) where targetcode = '31227061';

这不是我们想要的结果,具体改进法法如下:

 1 --方法一:对结果处理
 2 select max(datatime) datatime
 3       ,usercode
 4       ,deptcode
 5       ,deptname
 6       ,targetcode
 7       ,targetname
 8       ,sum(coalesce(day_value,0)) day_value
 9       ,sum(coalesce(mon_value,0)) mon_value
10       ,sum(coalesce(year_value,0)) year_value
11   from(
12 select datatime,usercode,deptcode,deptname,targetcode,targetname,day_value,mon_value,year_value
13   from MOVEBI.T_GMS_MBI_TARGET_DATA
14   pivot(max(targetvalue) for type in (1 day_value,2 mon_value,3 year_value)) where targetcode = '31227061')
15   group by usercode
16           ,deptcode
17           ,deptname
18           ,targetcode
19           ,targetname;
20 --方法二:对原始表处理
21 select *
22   from (select '20181017' datatime,
23                usercode,
24                deptcode,
25                deptname,
26                targetcode,
27                targetname,
28                targetvalue,
29                type
30           from MOVEBI.T_GMS_MBI_TARGET_DATA
31          where datatime in ('20181017', '201810')
32            and targetcode = '31227061') t
33            pivot(max(targetvalue) for type in (1 day_value,2 mon_value,3 year_value)) where targetcode = '31227061';

二、列转行unpivot

根据上面的例子创建tmp_2测试用表

select student,科目,成绩 from tmp_2 unpivot (成绩 for 科目 in (语文, 数学, 英语, 物理));

同样不使用unpivot也可以实现同样的效果,只是sql语句会很长,而且执行速度效率也没有前者高

select student,'语文' 科目, (select 语文 from tmp_2 where student=f.student) 成绩 from tmp_2 f
union
select student,'数学' 科目, (select 数学 from tmp_2 where student=f.student) 成绩 from tmp_2 f
union
select student,'英语' 科目, (select 英语 from tmp_2 where student=f.student) 成绩 from tmp_2 f
union
select student,'物理' 科目, (select 物理 from tmp_2 where student=f.student) 成绩 from tmp_2 f
-------------------------------------------
select student,'语文' 科目,语文 from tmp_2
union
select student,'数学' 科目,语文 from tmp_2
union
select student,'英语' 科目,语文 from tmp_2
union
select student,'物理' 科目,语文 from tmp_2

 

(注:此为学习记录笔记,仅供参考若有问题请指正,后续补充......)

参考文档:https://blog.csdn.net/xiaokui_wingfly/article/details/42419207

参考文档:https://www.cnblogs.com/harvey888/p/6735093.html

参考文档:https://www.cnblogs.com/markfeifei/p/4009343.html

posted @ 2018-11-23 17:16  Daily_Footprint  阅读(19990)  评论(0编辑  收藏  举报