數據庫查詢*分組排序取top n
要求：按照課程分組，查找每個課程最高的兩個成績。
數據文件如下：
第一列no為學號，第二列course為課程，第三列score為分數
mysql> select * from lesson;
+-------+---------+-------+
| no | course | score |
+-------+---------+-------+
| N0101 | Marth | 100 |
| N0102 | English | 12 |
| N0102 | Chinese | 55 |
| N0102 | History | 58 |
| N0102 | Marth | 25 |
| N0103 | English | 100 |
| N0103 | Chinese | 87 |
| N0103 | History | 88 |
| N0103 | Marth | 72 |
| N0104 | English | 20 |
| N0104 | Chinese | 60 |
| N0104 | History | 88 |
| N0104 | Marth | 56 |
| N0105 | English | 56 |
| N0105 | Chinese | 88 |
| N0105 | History | 88 |
| N0201 | English | 66 |
| N0201 | Chinese | 77 |
| N0201 | History | 80 |
| N0201 | Marth | 100 |
| N0202 | English | 35 |
| N0202 | Chinese | 56 |
| N0202 | History | 86 |
| N0202 | Marth | 99 |
| N0203 | English | 100 |
| N0203 | Chinese | 87 |
| N0203 | History | 88 |
| N0203 | Marth | 57 |
| N0204 | English | 98 |
| N0204 | Chinese | 100 |
| N0204 | History | 66 |
| N0204 | Marth | 71 |
| N0205 | English | 98 |
| N0205 | Chinese | 100 |
| N0205 | History | 66 |
| N0205 | Marth | 71 |
| N0301 | English | 66 |
| N0301 | Chinese | 89 |
| N0301 | History | 68 |
| N0301 | Marth | 83 |
| N0302 | English | 76 |
| N0302 | Chinese | 99 |
| N0302 | History | 80 |
| N0302 | Marth | 74 |
| N0303 | English | 100 |
| N0303 | Chinese | 100 |
| N0303 | History | 88 |
| N0303 | Marth | 57 |
| N0304 | English | 76 |
| N0304 | Chinese | 100 |
| N0304 | History | 66 |
| N0304 | Marth | 86 |
| N0305 | English | 98 |
| N0305 | Chinese | 100 |
| N0305 | History | 40 |
| N0305 | Marth | 59 |
| N0306 | English | 52 |
| N0306 | Chinese | 87 |
| N0306 | History | 72 |
| N0306 | Marth | 71 |
| N0101 | Chinese | 55 |
| N0101 | History | 84 |
| N0101 | English | 82 |
| N0101 | English | 82 |
+-------+---------+-------+
64 rows in set
在hive上查詢
select a.course,a.score from ( select course,score,row_number() over(partition by course order by score desc) as n from lesson )a where a.n<=2;
其中：
row_number() over(partition by course order by score desc)
意思是以課程分組，按成績遞減排序，並為每組中的數據打上行號的標記，從1開始。
這樣，再在外層套一層過濾行號小於等於2的即可:-D
原文：https://blog.csdn.net/wguangliang/article/details/50167283