oracle 分析函数之rank,dense_rank,row_number笔记

rank,dense_rank,row_number 函数都是为每条记录产生一个从1开始至N的自然数，N的值可能小于等于记录的总数。这3个函数的唯一区别在于，当碰到相同数据时的排名策略。

语法：rank() over(partition-clause order-by-clause)

rank:

rank函数返回一个唯一的值，除非遇到相同的数据时，此时所有相同数据的排名是一样的，同时会在最后一条相同记录和下一条记录排名之间空出排名。

dense_rank:

dense_rank函数返回一个唯一的值，除非当碰到相同数据时，此时所有相同数据的排名都是一样的。

row_number:

row_number函数返回一个唯一的值，当碰到相同数据时，排名按照记录集中记录的顺序依次递增。

建表如下：

create table SMALL_CUSTOMERS(CUSTOMER_ID NUMBER,SUM_ORDERS  NUMBER);     
insert into SMALL_CUSTOMERS (CUSTOMER_ID, SUM_ORDERS) values (1000, 10);     
insert into SMALL_CUSTOMERS (CUSTOMER_ID, SUM_ORDERS) values (1000, 20);     
insert into SMALL_CUSTOMERS (CUSTOMER_ID, SUM_ORDERS) values (1000, 30);     
insert into SMALL_CUSTOMERS (CUSTOMER_ID, SUM_ORDERS) values (800, 5);     
insert into SMALL_CUSTOMERS (CUSTOMER_ID, SUM_ORDERS) values (800, 10);     
insert into SMALL_CUSTOMERS (CUSTOMER_ID, SUM_ORDERS) values (800, 1); 

 

具体用法如下：

select customer_id,
       sum_orders,
       row_number() over(order by sum_orders desc) rowNumber,
       rank() over(order by sum_orders desc) rank,
       dense_rank() over(order by sum_orders desc) denseRank
  from small_customers t;

也可以进行分组：

select customer_id,
       sum_orders,
       row_number() over(partition by customer_id order by sum_orders desc) rowNumber,
       rank() over(partition by customer_id order by sum_orders desc) rank,
       dense_rank() over(partition by customer_id order by sum_orders desc) denseRank
  from small_customers t;