[SQL学习] 分析函数 min/max

- MIN函数的语法描述
- 从语法上看，有一个“OVER”类型的扩展，这便是分析函数语法部分
http://download.oracle.com/docs/cd/B19306_01/server.102/b14200/functions087.htm#SQLRF00667
MIN([ DISTINCT | ALL ] expr)
   [ OVER (analytic_clause) ]
- 有关分析函数的扩展可以参考Oracle的官方文档中的“Analytic Functions”描述：http://download.oracle.com/docs/cd/B19306_01/server.102/b14200/functions001.htm#SQLRF06174

- 通过实验看一下两种MIN函数的用法和区别
-- 创建测试表T，并初始化几条数据
create table t (group_id number(10), name varchar2(10), salary int);
insert into t values (1,'Tom',1200);
insert into t values (2,'Kary',2400);
insert into t values (2,'Joe',800);
insert into t values (3,'Erick',3600);
insert into t values (3,'Andy',600);
insert into t values (3,'Mary',300);
commit;

-- T表全貌
-- 共三组数据，group_id分别是1、2和3。第1组有一个人，第2组有两个人，第3组有三个人。最后一列是每个人的薪水值
select * from t;

  GROUP_ID NAME                               SALARY
---------- ------------------------------ ----------
         1 Tom                                  1200
         2 Kary                                 2400
         2 Joe                                   800
         3 Erick                                3600
         3 Andy                                  600
         3 Mary                                  300

6 rows selected.

-- 先看一下普通MIN函数的用法
-- 得到的最小值是按照“group_id, name, salary”三列进行分组后的每组最小值
select group_id, name, salary, min(salary) from t group by group_id, name, salary order by group_id;

  GROUP_ID NAME                               SALARY MIN(SALARY)
---------- ------------------------------ ---------- -----------
         1 Tom                                  1200        1200
         2 Joe                                   800         800
         2 Kary                                 2400        2400
         3 Andy                                  600         600
         3 Erick                                3600        3600
         3 Mary                                  300         300

6 rows selected.

-- 再看一下分析函数MIN得到的最小值
-- 注意这里得到的最小值与基本MIN函数得到的最小值区别，这里的最小值是每行记录所对应组内的最小薪酬值。这样的最小值是非常有意义的，在复杂的数据统计和分析系统中用处极大。

select group_id, name, salary, min(salary) over (partition by group_id) analytic_min from t order by group_id;

  GROUP_ID NAME                               SALARY ANALYTIC_MIN
---------- ------------------------------ ---------- ------------
         1 Tom                                  1200         1200
         2 Kary                                 2400          800
         2 Joe                                   800          800
         3 Erick                                3600          300
         3 Andy                                  600          300
         3 Mary                                  300          300

6 rows selected.