SQL - How to select a row having a column with max

2020-01-27 02:32发布

date                 value

18/5/2010, 1 pm        40
18/5/2010, 2 pm        20
18/5/2010, 3 pm        60
18/5/2010, 4 pm        30
18/5/2010, 5 pm        60
18/5/2010, 6 pm        25 

i need to query for the row having max(value)(i.e. 60). So, here we get two rows. From that, I need the row with the lowest time stamp for that day(i.e 18/5/2010, 3 pm -> 60)

标签: sql oracle
9条回答
兄弟一词,经得起流年.
2楼-- · 2020-01-27 02:48

Keywords like TOP, LIMIT, ROWNUM, ...etc are database dependent. Please read this article for more information.

http://en.wikipedia.org/wiki/Select_(SQL)#Result_limits

Oracle: ROWNUM could be used.

select * from (select * from table 
order by value desc, date_column) 
where rownum = 1;

Answering the question more specifically:

select high_val, my_key
from (select high_val, my_key
      from mytable
      where something = 'avalue'
      order by high_val desc)
where rownum <= 1
查看更多
我想做一个坏孩纸
3楼-- · 2020-01-27 02:54

In Oracle DB:

create table temp_test1 (id number, value number, description varchar2(20));

insert into temp_test1 values(1, 22, 'qq');
insert into temp_test1 values(2, 22, 'qq');
insert into temp_test1 values(3, 22, 'qq');
insert into temp_test1 values(4, 23, 'qq1');
insert into temp_test1 values(5, 23, 'qq1');
insert into temp_test1 values(6, 23, 'qq1');

SELECT MAX(id), value, description FROM temp_test1 GROUP BY value, description;

Result:
    MAX(ID) VALUE DESCRIPTION
    -------------------------
    6         23    qq1
    3         22    qq
查看更多
劫难
4楼-- · 2020-01-27 02:56

Technically, this is the same answer as @Sujee. It also depends on your version of Oracle as to whether it works. (I think this syntax was introduced in Oracle 12??)

SELECT *
FROM   table
ORDER BY value DESC, date_column ASC
FETCH  first 1 rows only;

As I say, if you look under the bonnet, I think this code is unpacked internally by the Oracle Optimizer to read like @Sujee's. However, I'm a sucker for pretty coding, and nesting select statements without a good reason does not qualify as beautiful!! :-P

查看更多
男人必须洒脱
5楼-- · 2020-01-27 02:56

The simplest answer would be

--Setup a test table called "t1"

create table t1
(date datetime,
value int)

-- Load the data. -- Note: date format different than in the question

insert into t1
Select '5/18/2010 13:00',40
union all
Select '5/18/2010 14:00',20
union all
Select '5/18/2010 15:00',60 
union all
Select '5/18/2010 16:00',30 
union all
Select '5/18/2010 17:00',60 
union all
Select '5/18/2010 18:00',25 

-- find the row with the max qty and min date.

select *
from t1
where value = 
    (select max(value)  from t1)
and date = 
    (select min(date) 
    from t1
    where value = (select max(value)  from t1))

I know you can do the "TOP 1" answer, but usually your solution gets just complicated enough that you can't use that for some reason.

查看更多
够拽才男人
6楼-- · 2020-01-27 02:59
SQL> create table t (mydate,value)
  2  as
  3  select to_date('18/5/2010, 1 pm','dd/mm/yyyy, hh am'), 40 from dual union all
  4  select to_date('18/5/2010, 2 pm','dd/mm/yyyy, hh am'), 20 from dual union all
  5  select to_date('18/5/2010, 3 pm','dd/mm/yyyy, hh am'), 60 from dual union all
  6  select to_date('18/5/2010, 4 pm','dd/mm/yyyy, hh am'), 30 from dual union all
  7  select to_date('18/5/2010, 5 pm','dd/mm/yyyy, hh am'), 60 from dual union all
  8  select to_date('18/5/2010, 6 pm','dd/mm/yyyy, hh am'), 25 from dual
  9  /

Table created.

SQL> select min(mydate) keep (dense_rank last order by value) mydate
  2       , max(value) value
  3    from t
  4  /

MYDATE                   VALUE
------------------- ----------
18-05-2010 15:00:00         60

1 row selected.

Regards, Rob.

查看更多
我只想做你的唯一
7楼-- · 2020-01-27 03:02

Answer is to add a having clause:

SELECT [columns]
FROM table t1
WHERE value= (select max(value) from table)
AND date = (select MIN(date) from table t2 where t1.value = t2.value)

this should work and gets rid of the neccesity of having an extra sub select in the date clause.

查看更多
登录 后发表回答