sum(sal) over (partition by deptno) --按照不同的部门分组，对部门自身的员工工资连续求和
partition by 是起着分组的作用
二、row_number over
row_number() over ([partition by col1] order by col2) ) as 别名
表示根据col1分组，在分组内部根据 col2排序
而这个“别名”的值就表示每组内部排序后的顺序编号（组内连续的唯一的）,[partition by col1] 可省略
row_number() over 按照partion by 分组，然后在每组内部按照 open_date 排序 得到返回的是内部排序的顺序编号
SELECT t.family_id,
t.package_id,
t.phone,
t.reason,
t.open_date,
ROW_NUMBER ()
OVER (PARTITION BY family_id, package_id, phone
ORDER BY open_date DESC)
rn
FROM zj_transaction_log t
WHERE TO_CHAR (t.open_date, 'yyyy-mm') >= '2011-03';
SELECT t.family_id,
t.package_id,
t.phone,
t.reason,
t.open_date,
ROW_NUMBER ()
OVER (PARTITION BY family_id, package_id, phone
ORDER BY package_id DESC)
rn
FROM zj_transaction_log t
WHERE TO_CHAR (t.open_date, 'yyyy-mm') >= '2011-03';
SELECT t.family_id,
t.package_id,
t.phone,
t.open_date,
t.reason,
COUNT( * )
OVER (PARTITION BY family_id, package_id, phone)
rn
FROM zj_transaction_log t
WHERE TO_CHAR (t.open_date, 'yyyy-mm') >= '2011-03';
SELECT t.family_id,
t.package_id,
t.phone,
t.open_date,
t.reason,
COUNT( * )
OVER (PARTITION BY family_id, package_id, phone
order by open_date)
rn
FROM zj_transaction_log t
WHERE TO_CHAR (t.open_date, 'yyyy-mm') >= '2011-03';
SELECT t.family_id,
t.package_id,
t.phone,
t.open_date,
t.reason,
COUNT( * )
OVER (PARTITION BY family_id, package_id, phone
order by package_id)
rn
FROM zj_transaction_log t
WHERE TO_CHAR (t.open_date, 'yyyy-mm') >= '2011-03';
SELECT t.family_id,
t.package_id,
t.phone,
t.reason,
COUNT( * )
OVER (PARTITION BY family_id, package_id, phone
ORDER BY open_date DESC
ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
rn
FROM zj_transaction_log t
WHERE TO_CHAR (t.open_date, 'yyyy-mm') >= '2011-03';
SELECT t.family_id,
t.package_id,
t.phone,
t.reason,
COUNT( * )
OVER (PARTITION BY family_id, package_id, phone
ORDER BY package_id DESC
ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
rn
FROM zj_transaction_log t
WHERE TO_CHAR (t.open_date, 'yyyy-mm') >= '2011-03';
rank 的分析功能语法：
RANK ( ) OVER ( [query_partition_clause] order_by_clause )
rank 的合计功能语法：
RANK ( expr [, expr]... ) WITHIN GROUP
( ORDER BY
expr [ DESC | ASC ] [NULLS { FIRST | LAST }]
[, expr [ DESC | ASC ] [NULLS { FIRST | LAST }]]...
)
对于分析功能，
SELECT a.*,RANK() OVER(PARTITION BY col2 ORDER BY col1) "Rank" FROM table a;
已col2分组col1排序，rank与dense_rank用法相当，区别如下：
例如：当rank时为：
SELECT f.phone, f.student_name,
RANK () OVER (PARTITION BY f.phone ORDER BY f.student_name) cnt
FROM zs_family f
WHERE NVL (f.is_test, 0) <> 1
AND LENGTH (f.phone) = 11
AND f.phone NOT LIKE '%0%'
而如果用dense_rank时为：
SELECT f.phone, f.student_name,
dense_RANK () OVER (PARTITION BY f.phone ORDER BY f.student_name) cnt
FROM zs_family f
WHERE NVL (f.is_test, 0) <> 1
AND LENGTH (f.phone) = 11
AND f.phone NOT LIKE '%0%'
SELECT RANK ('********188','陈雅琳') within group (ORDER BY f.phone,f.student_name) cnt
FROM zs_family f
WHERE NVL (f.is_test, 0) <> 1
AND LENGTH (f.phone) = 11
AND f.phone NOT LIKE '%0%';
cnt
258
感觉就是查询定位该记录的在表中位置
五、max\min over
功能：生成2列获得字段最大值、最小值
例如：
查询员工信息与其中工资最高最低员工 select ename,sal,max(sal) over(), min(sal) over() from emp;
六、lag\lead over
功能：
得到当前行上一行或者下一行的数据
例子：
要求用一个SQL语句，统计每个月及上月和下月的订单量
要求列表输出为
月份 当月订单量 上月订单量 下月订单量
select to_char(d.update_time, 'yyyymm'),
count(code),
lag(count(code)) over(order by count(code)) 上月订单量,
lead(count(code)) over(order by count(code)) 下有订单量
from V_NWRZ_dwxx_TEST2 d
group by to_char(d.update_time, 'yyyymm');