求助 oracle 一条sql 语句, 急用可以加分

likaiyihou512 2013-09-22 10:42:03

有表如下：
字段 X Y Z A
x 10 1 2
x 10 2 3
x 11 1 4
x 11 2 6
y 15 5 1
y 15 6 4
y 16 5 2
y 16 6 8
求：按字段X 分组，取 Y 最大的记录，
例如要得到的结果是：
x 11 4 6,
y 16 2 8

...全文

516 22 打赏收藏转发到动态举报

写回复

用AI写文章

22 条回复

切换为时间正序

请发表友善的回复…

发表回复

likaiyihou512 2013-09-27

打赏
举报

试下结贴能回复么

likaiyihou512 2013-09-27

打赏
举报


create or replace view se_mark_examination_v as
with a as (select se.sys_user_id,se_exam_p.Get_Employee_Names(se.sys_user_id) as SYS_USER_NAME,se.se_paper_id,sp.paper_name,
       max(sea.attribute_10) as exam_num
from   se_examinee_answer sea,se_paper sp ,se_question sq,se_examinee se
where sea.se_examinee_id = se.id and sea.se_question_id = sq.id  and sp.id = se.se_paper_id
      and se.status = '3' and sp.reply_num is not null and sp.reply_grade is not null and sea.attribute_10 is not null
group by se.sys_user_id,se_exam_p.Get_Employee_Names(se.sys_user_id),se.se_paper_id,sp.paper_name),

b as (select se.sys_user_id,se_exam_p.Get_Employee_Names(se.sys_user_id) as SYS_USER_NAME,se.se_paper_id,sp.paper_name,
       sq.question_type,sum(sea.answer_of_score) as scores,sea.attribute_10
from   se_examinee_answer sea,se_paper sp ,se_question sq,se_examinee se,a
where sea.se_examinee_id = se.id and sea.se_question_id = sq.id  and sp.id = se.se_paper_id
      and a.sys_user_id = se.sys_user_id and a.se_paper_id = se.se_paper_id and a.exam_num = sea.attribute_10
      and se.status = '3' and sp.reply_num is not null and sp.reply_grade is not null and sea.attribute_10 is not null
group by se.sys_user_id,se_exam_p.Get_Employee_Names(se.sys_user_id),se.se_paper_id,sp.paper_name,sq.question_type,sea.attribute_10
)

select b.sys_user_id,b.SYS_USER_NAME,s.id as se_examinee_id,b.se_paper_id,b.paper_name,b.attribute_10 as exam_num,s.real_end_time as exam_start_time
       ,max(decode(b.question_type, '1', scores,0)) dx_scores1,--单选
       max(decode(b.question_type, '2', scores,0)) duox_scores2,--多选
       max(decode(b.question_type, '3', scores,0)) tk_scores3,--填空
       max(decode(b.question_type, '4', scores,0)) pd_scores4,--判断题
       max(decode(b.question_type, '5', scores,0)) jd_scores5--简单题
from b
left join se_examinee s on b.se_paper_id = s.se_paper_id and b.sys_user_id =s.sys_user_id
group by b.sys_user_id,b.SYS_USER_NAME,s.id,b.se_paper_id,b.paper_name,b.attribute_10,s.real_end_time
order by b.sys_user_id,b.se_paper_id,b.attribute_10

解决了，感谢大家的回答，及给予的思路，赋上解决sql给网友参考

likaiyihou512 2013-09-22

打赏
举报

引用 6 楼 kuaileyangle 的回复:

最近学了个函数，来抛砖引玉一下：



WITH T AS

 (SELECT 'x' X, 10 Y, 1 Z, 2 A

    FROM DUAL

  UNION ALL

  SELECT 'x', 10, 2, 3

    FROM DUAL

  UNION ALL

  SELECT 'x', 11, 1, 4

    FROM DUAL

  UNION ALL

  SELECT 'x', 11, 2, 6

    FROM DUAL

  UNION ALL

  SELECT 'y', 15, 5, 1

    FROM DUAL

  UNION ALL

  SELECT 'y', 15, 6, 4

    FROM DUAL

  UNION ALL

  SELECT 'y', 16, 5, 2

    FROM DUAL

  UNION ALL

  SELECT 'y', 16, 6, 8 FROM DUAL)

SELECT X, MAX(Y), WMSYS.WM_CONCAT(A)

  FROM (SELECT * FROM T WHERE Y IN (SELECT MAX(Y) FROM T GROUP BY X))

 GROUP BY X;

接近楼主的要求了，但还不完美！

WMSYS.WM_CONCAT 开始也考虑了用这个内部函数，但是弄成了一个字段，前台显示不好看。实在没有其他的方法的话考虑用这个函数

likaiyihou512 2013-09-22

打赏
举报

引用 5 楼 lyliu602 的回复:

你用decode转换一下吧
decode(z,1,a)a
decode(z,2,a)b

select t.x,t.y,decode(t.z, '1', t.a) as A , decode(t.z, '2', t.a) as B
from (
select x,y,z,a from test where (x,y) in
(select x,max(to_number(y)) from test group by x)
--group by x,y
order by x,y ) t -- ,wmsys.wm_concat(A)

现在 2013-09-22

打赏
举报

最近学了个函数，来抛砖引玉一下：


WITH T AS
 (SELECT 'x' X, 10 Y, 1 Z, 2 A
    FROM DUAL
  UNION ALL
  SELECT 'x', 10, 2, 3
    FROM DUAL
  UNION ALL
  SELECT 'x', 11, 1, 4
    FROM DUAL
  UNION ALL
  SELECT 'x', 11, 2, 6
    FROM DUAL
  UNION ALL
  SELECT 'y', 15, 5, 1
    FROM DUAL
  UNION ALL
  SELECT 'y', 15, 6, 4
    FROM DUAL
  UNION ALL
  SELECT 'y', 16, 5, 2
    FROM DUAL
  UNION ALL
  SELECT 'y', 16, 6, 8 FROM DUAL)
SELECT X, MAX(Y), WMSYS.WM_CONCAT(A)
  FROM (SELECT * FROM T WHERE Y IN (SELECT MAX(Y) FROM T GROUP BY X))
 GROUP BY X;

接近楼主的要求了，但还不完美！

无敌小二傻 2013-09-22

打赏
举报

你用decode转换一下吧 decode(z,1,a)a decode(z,2,a)b

likaiyihou512 2013-09-22

打赏
举报

引用 2 楼 lyliu602 的回复:

看看懂你的z列是不是没有用到还是写错了呢？ select x,max(y),z,a group by x,z,a

我取得是Z列不同， x 11 1 4 x 11 2 6 你看着两列，我要得到：x 11 4 6

请叫我-雷人 2013-09-22

打赏
举报

数据确实乱，不过语法大体和楼上一致

无敌小二傻 2013-09-22

打赏
举报

看看懂你的z列是不是没有用到还是写错了呢？ select x,max(y),z,a group by x,z,a

likaiyihou512 2013-09-22

打赏
举报

自己顶一个，来个大神，给解决下

gaojiebao123 2013-09-22

打赏
举报

在实际中有什么应用能说下不？

无敌小二傻 2013-09-22

打赏
举报

引用 7 楼 likaiyihou512 的回复:

[quote=引用 5 楼 lyliu602 的回复:] 你用decode转换一下吧 decode(z,1,a)a decode(z,2,a)b

select t.x,t.y,decode(t.z, '1', t.a) as A , decode(t.z, '2', t.a) as B from ( select x,y,z,a from test where (x,y) in (select x,max(to_number(y)) from test group by x) --group by x,y order by x,y ) t -- ,wmsys.wm_concat(A)

[/quote] select t.x,t.y,decode(t.z, '1', t.a) as A , decode(t.z, '2', t.a) as B from ( 改为： select t.x,t.y,decode(t.z, '1', t.a,null,t.a) as A , decode(t.z, '2', t.a,null,t.a) as B from (

u010412956 2013-09-22

打赏
举报

引用 13 楼 likaiyihou512 的回复:

Z 字段来控制（最多五种），有值得就显示值，没有的填零

最多五种的话，就这样写


WITH T AS
 (SELECT 'x' X, 10 Y, 1 Z, 2 A
    FROM DUAL
  UNION ALL
  SELECT 'x', 10, 2, 3
    FROM DUAL
  UNION ALL
  SELECT 'x', 11, 1, 4
    FROM DUAL
  UNION ALL
  SELECT 'x', 11, 2, 6
    FROM DUAL
  UNION ALL
  SELECT 'y', 15, 5, 1
    FROM DUAL
  UNION ALL
  SELECT 'y', 15, 6, 4
    FROM DUAL
  UNION ALL
  SELECT 'y', 16, 5, 2
    FROM DUAL
  UNION ALL
  SELECT 'y', 16, 6, 8
    FROM DUAL
  UNION ALL
  SELECT 'y', 16, 7, 9 FROM DUAL),
t1 as
 (select x, y, a, rn
    from (select x,
                 y,
                 a,
                 dense_rank() over(partition by x order by y desc) dr,
                 row_number() over(partition by x order by y desc) rn
            from t)
   where dr = 1)
select * from t1 pivot(max(a) for rn in(1, 2, 3, 4, 5))

needforskill 2013-09-22

打赏
举报

select x,gy,a1,a2,a3,a4,a5 from
(
select x,gy
,lag(a,case when rb-1 >=0 then rb-1 else 999 end,0) over(partition by x,gy order by a) a1
,lag(a,case when rb-2 >=0 then rb-2 else 999 end,0) over(partition by x,gy order by a) a2
,lag(a,case when rb-3 >=0 then rb-3 else 999 end,0) over(partition by x,gy order by a) a3
,lag(a,case when rb-4 >=0 then rb-4 else 999 end,0) over(partition by x,gy order by a) a4
,lag(a,case when rb-5 >=0 then rb-5 else 999 end,0) over(partition by x,gy order by a) a5
,row_number() over( partition by x order by  rownum desc) xrb
from
(select X,y,max(Y) over(partition by x) gy,
 rank() over (partition by x,y order by a asc) rb
,a
from
(select 'x' X, 10 Y,1 Z,2 A from dual
union
select 'x' X, 10 Y,2 Z,3 A from dual
union
select 'x' X, 11 Y,1 Z,4 A from dual
union
select 'x' X, 11 Y,2 Z,6 A from dual
union
select 'x' X, 11 Y,1 Z,7 A from dual
union
select 'x' X, 11 Y,2 Z,8 A from dual
union
select 'x' X, 11 Y,2 Z,9 A from dual
union
select 'y' X, 15 Y,5 Z,1 A from dual
union
select 'y' X, 15 Y,6 Z,4 A from dual
union
select 'y' X, 16 Y,5 Z,2 A from dual
union
select 'y' X, 16 Y,5 Z,9 A from dual
union
select 'y' X, 16 Y,6 Z,8 A from dual))
where y=gy)
where xrb=1

这个少了嵌套,找不到的列置为0

needforskill 2013-09-22

打赏
举报

select x,gy,a1,a2,a3,a4,a5 from
(
select x,gy
,lag(a,case when rb-1 >=0 then rb-1 else 999 end) over(partition by x,gy order by a) a1
,lag(a,case when rb-2 >=0 then rb-2 else 999 end) over(partition by x,gy order by a) a2
,lag(a,case when rb-3 >=0 then rb-3 else 999 end) over(partition by x,gy order by a) a3
,lag(a,case when rb-4 >=0 then rb-4 else 999 end) over(partition by x,gy order by a) a4
,lag(a,case when rb-5 >=0 then rb-5 else 999 end) over(partition by x,gy order by a) a5
,row_number() over( partition by x order by  rownum desc) xrb
from (select
x,gy,a,row_number() over(partition by x order by a asc) rb
from
(select X,y,max(Y) over(partition by Z) gy
,a
from
(select 'x' X, 10 Y,1 Z,2 A from dual
union
select 'x' X, 10 Y,2 Z,3 A from dual
union
select 'x' X, 11 Y,1 Z,4 A from dual
union
select 'x' X, 11 Y,2 Z,6 A from dual
union
select 'x' X, 11 Y,1 Z,7 A from dual
union
select 'x' X, 11 Y,2 Z,8 A from dual
union
select 'x' X, 11 Y,2 Z,9 A from dual
union
select 'y' X, 15 Y,5 Z,1 A from dual
union
select 'y' X, 15 Y,6 Z,4 A from dual
union
select 'y' X, 16 Y,5 Z,2 A from dual
union
select 'y' X, 16 Y,6 Z,8 A from dual))
where y=gy))
where xrb=1

用的分析函数有点多，而且还是嵌套,如果数据量大(千万级)的话，建议分几步作

likaiyihou512 2013-09-22

打赏
举报

引用 14 楼 qq304213346 的回复:

with t as( select 'x' X, 10 Y,1 Z,2 A from dual union select 'x' X, 10 Y,2 Z,3 A from dual union select 'x' X, 11 Y,1 Z,4 A from dual union select 'x' X, 11 Y,2 Z,6 A from dual union select 'y' X, 15 Y,5 Z,1 A from dual union select 'y' X, 15 Y,6 Z,4 A from dual union select 'y' X, 16 Y,5 Z,2 A from dual union select 'y' X, 16 Y,6 Z,8 A from dual) select x,y, wmsys.wm_concat(a) from t where (x,y) in ( select x,max(y) from t group by x) group by x,y

如果A列有存在为null 的数据，使用这个函数就不好使了

请叫我-雷人 2013-09-22

打赏
举报

with t as( select 'x' X, 10 Y,1 Z,2 A from dual union select 'x' X, 10 Y,2 Z,3 A from dual union select 'x' X, 11 Y,1 Z,4 A from dual union select 'x' X, 11 Y,2 Z,6 A from dual union select 'y' X, 15 Y,5 Z,1 A from dual union select 'y' X, 15 Y,6 Z,4 A from dual union select 'y' X, 16 Y,5 Z,2 A from dual union select 'y' X, 16 Y,6 Z,8 A from dual) select x,y, wmsys.wm_concat(a) from t where (x,y) in ( select x,max(y) from t group by x) group by x,y

likaiyihou512 2013-09-22