row_number( )、rank( )、,DENSE_RANK( )、ntile( )

row_number( )、rank( )、,DENSE_RANK( )、ntile( )下面以实例分别简单讲解。

1 .row_number( )
         先来点数据,先建个表

CREATE   TABLE  Person(
FirstName  VARCHAR ( 10 ),
Age  INT ,
Gender  CHAR ( 1 ))
INSERT   INTO  Person  VALUES  ( ' Ted ' , 23 , ' M ' )
INSERT   INTO  Person  VALUES  ( ' John ' , 40 , ' M ' )
INSERT   INTO  Person  VALUES  ( ' George ' , 6 , ' M ' )
INSERT   INTO  Person  VALUES  ( ' Mary ' , 11 , ' F ' )
INSERT   INTO  Person  VALUES  ( ' Sam ' , 17 , ' M ' )
INSERT   INTO  Person  VALUES  ( ' Doris ' , 6 , ' F ' )
INSERT   INTO  Person  VALUES  ( ' Frank ' , 38 , ' M ' )
INSERT   INTO  Person  VALUES  ( ' Larry ' , 5 , ' M ' )
INSERT   INTO  Person  VALUES  ( ' Sue ' , 29 , ' F ' )
INSERT   INTO  Person  VALUES  ( ' Sherry ' , 11 , ' F ' )
INSERT   INTO  Person  VALUES  ( ' Marty ' , 23 , ' F ' )
ROW_NUMBER()  OVER  ( ORDER   BY  Age)  AS   [ Row Number by Age ] ,
FirstName,
Age
FROM  Person

-- ------------------------                 ----------            --------
1                                                 Larry                    5
2                                                 Doris                    6
3                                                 George                6
4                                                 Mary                    11
5                                                 Sherry                  11
6                                                 Sam                     17
7                                                 Ted                      23
8                                                 Marty                    23
9                                                 Sue                      29
10                                               Frank                   38
11                                               John                     40

与sql server2000对比:
如果在sql server2000中实现相对麻烦一些，我们可以利用IDENTITY()函数实现，但IDENTITY()函数只能用在sql server2000临时表中，因此需要将数据检索到临时表里。
select   identity ( int , 1 , 1 )  as   [ Row Number by Age ] ,FirstName,Age  into  #A  from  Person  order   by  Age
select   *   from  #A
drop   table  #a

如果不想按年龄排序,可以这样写
SELECT  ROW_NUMBER()  OVER  ( ORDER   BY  ( SELECT   1 ))  AS   [ Row Number by Record Set ] ,
FirstName,
Age
FROM  Person

另外一个例子
SELECT  ROW_NUMBER()  OVER  (PARTITION  BY  Gender  ORDER   BY  Age)  AS   [ Partition by Gender ] ,
FirstName,
Age,
Gender
FROM  Person

注意,姓名M开始,序号又从1, 2 ,3开始了

 

2 .RANK( )函数
         先看例子
SELECT  RANK()  OVER  ( ORDER   BY  Age)  AS   [ Rank by Age ] ,
FirstName,
Age
FROM  Person

输出如下:
Rank  by  Age                 FirstName         Age
-- ------------------ ---------- -----------
1                            Larry              5
2                            Doris              6
2                            George             6
4                            Mary               11
4                            Sherry             11
6                            Sam                17
7                            Ted                23
7                            Marty              23
9                            Sue                29
10                           Frank              38
11                           John               40

看到了么,同年岭的话,将有相同的顺序,顺序成1, 2 , 2 ,4了。

与sql server2000对比:
出现了RANK()函数实在是方便,在sql server2000里实现排序并列的问题麻烦很多。
select   [ Rank by Age ] = isnull (( select   count ( * )  from  person  where  Age > A.Age), 0 ) + 1 ,FirstName,Age  from  Person A  order   by   [ Rank by Age ]

 

SELECT  RANK()  OVER (PARTITION  BY  Gender  ORDER   BY  Age)  AS   [ Partition by Gender ] ,
FirstName, Age, Gender  FROM  Person输出为

Partition  by  Gender         FirstName         Age                Gender
-- ------------------ ---------- ----------- ------
1                            Doris              6                   F
2                            Mary               11                  F
2                            Sherry             11                  F
4                            Sue                29                  F
1                            Larry              5                   M
2                            George             6                   M
3                            Sam                17                  M
4                            Ted                23                  M
4                            Marty              23                  M
6                            Frank              38                  M
7                            John               40                  M

可以看到,按性别分组了,每个性别分组里,继续是用了rank( )函数

3 .DENSE_RANK( )函数
          SELECT  DENSE_RANK()  OVER  ( ORDER   BY  Age)  AS   [ Dense Rank by Age ] ,
         FirstName,
         Age
          FROM  Person

输出结果为:
Dense Rank  by  Age          FirstName        Age
-- ------------------ ---------- -----------
1                           Larry             5
2                           Doris             6
2                           George            6
3                           Mary              11
3                           Sherry            11
4                           Sam               17
5                           Ted               23
5                           Marty             23
6                           Sue               29
7                           Frank             38
8                           John              40

看到了么,和rank函数区别是,顺序始终是连续的,Doris 和George同年,都是排第2位,但之后的mary不象rank函数那样排第4,而是排第3位了


4 .ntile( )函数
SELECT  FirstName,
Age,
NTILE( 3 )  OVER  ( ORDER   BY  Age)  AS   [ Age Groups ]
FROM  Person

输出结果：
FirstName        Age               Age Groups
-- -------- ----------- --------------------
Larry                 5                    1
Doris                 6                    1
George             6                    1
Mary                 11                  1
Sherry              11                   2
Sam                 17                   2
Ted                  23                   2
Marty               23                   2
Sue                 29                   3
Frank              38                   3
John                40                   3
这个函数按照ntile(n)中的N,把记录强制分成多少段,11条记录现在分成3段了,lary到mary是第1段,sherry到maty是第2段,sue到john是第3段