分析(窗口)函数：NTH_VALUE

NTH_VALUE 返回 analytic_clause 定义的窗口中第 n 行的 measure_expr 值。返回的值具有 measure_expr 的数据类型。

NTH_VALUE (measure_expr, n) [ FROM { FIRST | LAST } ][ { RESPECT | IGNORE } NULLS ] OVER (analytic_clause)

参数：

实操演练：

-- 建表
CREATE TABLE emp_msg(deptno INT, ename VARCHAR(30),sal INT, MGR VARCHAR(30), hiredate VARCHAR(50));

-- 插入数据
INSERT INTO emp_msg VALUES(10,'CLARK', 2750, 7839, '2018-05-01');       
INSERT INTO emp_msg VALUES(10,'KING', 5300, NULL, '2018-05-10');       
INSERT INTO emp_msg VALUES(10,'MILLER', 1600, 7782, '2018-06-01');        
INSERT INTO emp_msg VALUES(20,'ADAMS', 1400, 7788, '2018-05-21');       
INSERT INTO emp_msg VALUES(20,'FORD', 3300, 7566, '2018-06-01');      
INSERT INTO emp_msg VALUES(20,'JONES', 3275, 7839, '2018-06-20');      
INSERT INTO emp_msg VALUES(20,'SCOTT', 3300, 7566, '2018-07-01');    
INSERT INTO emp_msg VALUES(20,'SMITH', 1100, 7902, '2018-07-10');   
INSERT INTO emp_msg VALUES(30,'ALLEN', 1900, 7698, '2018-08-05'); 
INSERT INTO emp_msg VALUES(30,'BLAKE', 3150, 7839, '2018-06-10'); 
INSERT INTO emp_msg VALUES(30,'JAMES', 1250, 7698, '2018-09-05');    
INSERT INTO emp_msg VALUES(30,'MARTIN', 1550, 7698, '2018-10-01');
INSERT INTO emp_msg VALUES(30,'TURNER', 1800, 7698, '2019-05-01');
INSERT INTO emp_msg VALUES(30,'WARD', 1550, 7698, '2019-05-10');

-- 查看表
select * FROM emp_msg;

-- 按部门 deptno 分组，查询每个部门的人员的薪资和该部门内排名第三的薪资金额的对比
-- 执行以下语句：
--  ,rank()over(PARTITION BY deptno ORDER BY sal DESC) as ranking

SELECT  deptno
       ,ename
	   ,sal
	   ,nth_value(sal, 3)OVER (PARTITION BY deptno ORDER BY sal DESC 
rows BETWEEN unbounded preceding AND unbounded following)  AS third_most_sal 
FROM emp_msg ORDER BY deptno,sal DESC;

输出结果：

+--------+--------+------+----------------+
| DEPTNO | ENAME  | SAL  | THIRD_MOST_SAL |
+--------+--------+------+----------------+
|     10 | KING   | 5300 |           1600 |
|     10 | CLARK  | 2750 |           1600 |
|     10 | MILLER | 1600 |           1600 |
|     20 | FORD   | 3300 |           3275 |
|     20 | SCOTT  | 3300 |           3275 |
|     20 | JONES  | 3275 |           3275 |
|     20 | ADAMS  | 1400 |           3275 |
|     20 | SMITH  | 1100 |           3275 |
|     30 | BLAKE  | 3150 |           1800 |
|     30 | ALLEN  | 1900 |           1800 |
|     30 | TURNER | 1800 |           1800 |
|     30 | MARTIN | 1550 |           1800 |
|     30 | WARD   | 1550 |           1800 |
|     30 | JAMES  | 1250 |           1800 |
+--------+--------+------+----------------+

附图：

按部门 deptno 分组，查询每个部门的人员的薪资和该部门内排名第三的薪资金额的对比

代码变式：

SELECT  deptno
       ,ename
	   ,sal
	   ,nth_value(sal, 3)OVER (PARTITION BY deptno ORDER BY sal DESC )  AS third_most_sal 
FROM emp_msg ORDER BY deptno,sal DESC;

输出结果：
（默认为当前行，所以输出为空值），这点请注意，这一块的话，可以与上者进行对比；