group fetch top N Records
Often see the problem, how to take out the first N records of each group. Convenient for everyone to refer to the common several solutions listed below.
question: The table below, requires the removal of the top two classes (allow second)
Table1
+----+------+------+-----+
| ID | SName | Clsno | score|
+----+------+------+-----+
| 1 | AAAA | C1 | 67 |
| 2 | BBBB | C1 | 55 |
| 3 | CCCC | C1 | 67 |
| 4 | DDDD | C1 | 65 |
| 5 | eeee | C1 | 95 |
| 6 | FFFF | C2 | 57 |
| 7 | GGGG | C2 | 87 |
| 8 | HHHH | C2 | 74 |
| 9 | IIII | C2 | 52 |
| 10 | JJJJ | C2 | 81 |
| 11 | KKKK | C2 | 67 |
| 12 | llll | C2 | 66 |
| 13 | MMMM | C2 | 63 |
| 14 | NNNN | C3 | 99 |
| 15 | oooo | C3 | 50 |
| 16 | PPPP | C3 | 59 |
| 17 | QQQQ | C3 | 66 |
| 18 | RRRR | C3 | 76 |
| 19 | SSSS | C3 | 50 |
| 20 | TTTT | C3 | 50 |
| 21 | uuuu | C3 | 64 |
| 22 | VVVV | C3 | 74 |
+----+------+------+-----+
The results are as follows
+----+------+------+-----+
| ID | SName | Clsno | score|
+----+------+------+-----+
| 5 | eeee | C1 | 95 |
| 1 | AAAA | C1 | 67 |
| 3 | CCCC | C1 | 67 |
| 7 | GGGG | C2 | 87 |
| 10 | JJJJ | C2 | 81 |
| 14 | NNNN | C3 | 99 |
| 18 | RRRR | C3 | 76 |
+----+------+------+-----+
Method One:
Select A.id,a.sname,a.clsno,a.score
From Table1 a LEFT join Table1 B on A.clsno=b.clsno and A.score<b.score
GROUP BY A.id,a.sname,a.clsno,a.score
Having count (b.id) <2
ORDER BY A.clsno,a.score Desc
Method Two:
SELECT *
From Table1 A
where 2> (select COUNT (*) from Table1 where Clsno=a.clsno and Score>a.score)
ORDER BY A.clsno,a.score Desc
Method Three:
SELECT *
From Table1 A
where ID in (select ID from Table1 where clsno=a.clsno order BY score desc limit 2)
ORDER BY A.clsno,a.score Desc
Method....
Here are a list of implementations of various SQL statements, some of which are MySQL-specific (Limit, other databases can be based on actual changes, such as Oracle's ROWNUM,MS SQL SERVER top,..), sometimes supported by the SQL standard. But the efficiency and application of the occasion may be different. Specific application can be based on the actual table records, the index of the situation to choose.
Special case N=1, that is, take the largest/smallest record.
+----+------+------+-----+
| ID | SName | Clsno | score|
+----+------+------+-----+
| 5 | eeee | C1 | 95 |
| 7 | GGGG | C2 | 87 |
| 14 | NNNN | C3 | 99 |
+----+------+------+-----+
SELECT *
From Table1 A
Where NOT EXISTS (select 1 from Table1 where Clsno=a.clsno and Score>a.score);
Select A.*
From Table1 a inner joins (select Clsno, Max (score) as Mscore from Table1 Group by Clsno) b
On A.clsno=b.clsno and A.score=b.score
SELECT *
From (SELECT * from Table1 ORDER BY score desc) T
GROUP BY Clsno
Group fetch top N Records