如何根据另一列计算以下n行的平均值 - SQL（Oracle）

Question

我正在尝试按月计算每个 POLICY_ID 的平均月保费价值，如下所示。当客户将他/她的年度付款频率更新为不同于 12 的值时，我需要手动计算 PREMIUM 的平均每月值。我怎样才能达到 MONTHLY _PREMIUM_DESIRED 中显示的值？ 提前致谢。

注意：Oracle 版本 12c

我尝试过的：

SELECT 
    T.*,
    SUM(PREMIUM) OVER(PARTITION BY T.POLICY_ID ORDER BY T.POLICY_ID, T.PAYMENT_DATE ROWS BETWEEN CURRENT ROW AND 12/T.YEARLY_PAYMENT_FREQ-1 FOLLOWING ) / (12/T.YEARLY_PAYMENT_FREQ) MONTLY_PREMIUM_CALCULATED
FROM MYTABLE T
;

数据代码：

DROP TABLE MYTABLE;
CREATE TABLE MYTABLE (POLICY_ID NUMBER(11), PAYMENT_DATE DATE, PREMIUM NUMBER(5), YEARLY_PAYMENT_FREQ NUMBER(2),MONTHLY_PREMIUM_DESIRED NUMBER(5));                                  
INSERT INTO MYTABLE VALUES (1, DATE '2014-10-01',120,12,120);
INSERT INTO MYTABLE VALUES (1, DATE  '2014-11-01',360,4,120);
INSERT INTO MYTABLE VALUES (1, DATE '2014-12-01',0,4,120);
INSERT INTO MYTABLE VALUES (1, DATE '2015-01-01',0,4,120);
INSERT INTO MYTABLE VALUES (1, DATE '2015-02-01',360,4,120);
INSERT INTO MYTABLE VALUES (1, DATE '2015-03-01',0,4,120);
INSERT INTO MYTABLE VALUES (1, DATE '2015-04-01',0,4,120);
INSERT INTO MYTABLE VALUES (1, DATE '2015-05-01',720,2,120);
INSERT INTO MYTABLE VALUES (1, DATE '2015-06-01',0,2,120);
INSERT INTO MYTABLE VALUES (1, DATE '2015-07-01',0,2,120);
INSERT INTO MYTABLE VALUES (1, DATE '2015-08-01',0,2,120);
INSERT INTO MYTABLE VALUES (1, DATE '2015-09-01',0,2,120);
INSERT INTO MYTABLE VALUES (1, DATE '2015-10-01',0,2,120);
INSERT INTO MYTABLE VALUES (1, DATE '2015-11-01',120,12,120);
INSERT INTO MYTABLE VALUES (2, DATE '2015-01-01',60,3,15);
INSERT INTO MYTABLE VALUES (2, DATE '2015-02-01',0,3,15);
INSERT INTO MYTABLE VALUES (2, DATE '2015-03-01',0,3,15);
INSERT INTO MYTABLE VALUES (2, DATE '2015-04-01',0,3,15);
INSERT INTO MYTABLE VALUES (2, DATE '2015-05-01',180,1,15);
INSERT INTO MYTABLE VALUES (2, DATE '2015-06-01',0,1,15);
INSERT INTO MYTABLE VALUES (2, DATE '2015-07-01',0,1,15);
INSERT INTO MYTABLE VALUES (2, DATE '2015-08-01',0,1,15);
INSERT INTO MYTABLE VALUES (2, DATE '2015-09-01',0,1,15);
INSERT INTO MYTABLE VALUES (2, DATE '2015-10-01',0,1,15);
INSERT INTO MYTABLE VALUES (2, DATE '2015-11-01',0,1,15);
INSERT INTO MYTABLE VALUES (2, DATE '2015-12-01',0,1,15);
INSERT INTO MYTABLE VALUES (2, DATE '2016-01-01',0,1,15);
INSERT INTO MYTABLE VALUES (2, DATE '2016-02-01',0,1,15);
INSERT INTO MYTABLE VALUES (2, DATE '2016-03-01',0,1,15);
INSERT INTO MYTABLE VALUES (2, DATE '2016-04-01',0,1,15);
INSERT INTO MYTABLE VALUES (2, DATE '2016-05-01',15,12,15);
INSERT INTO MYTABLE VALUES (2, DATE '2016-06-01',15,12,15);
SELECT * FROM MYTABLE;

编辑：

无论付款频率如何，客户也可以更改溢价金额。下面，对于 POLICY_ID = 1，我添加了从“2015/11/01”开始的新记录来演示这种情况。在这种情况下，平均每月保费从 120 增加到 240。 还删除了屏幕截图以使问题更具可读性。

DROP TABLE MYTABLE2;
CREATE TABLE MYTABLE2 (POLICY_ID NUMBER(11), PAYMENT_DATE DATE, PREMIUM NUMBER(5), YEARLY_PAYMENT_FREQ NUMBER(2),MONTHLY_PREMIUM_DESIRED NUMBER(5));
INSERT INTO MYTABLE2 VALUES (1, DATE '2014-10-01',120,12,120);
INSERT INTO MYTABLE2 VALUES (1, DATE  '2014-11-01',360,4,120);
INSERT INTO MYTABLE2 VALUES (1, DATE '2014-12-01',0,4,120);
INSERT INTO MYTABLE2 VALUES (1, DATE '2015-01-01',0,4,120);
INSERT INTO MYTABLE2 VALUES (1, DATE '2015-02-01',360,4,120);
INSERT INTO MYTABLE2 VALUES (1, DATE '2015-03-01',0,4,120);
INSERT INTO MYTABLE2 VALUES (1, DATE '2015-04-01',0,4,120);
INSERT INTO MYTABLE2 VALUES (1, DATE '2015-05-01',720,2,120);
INSERT INTO MYTABLE2 VALUES (1, DATE '2015-06-01',0,2,120);
INSERT INTO MYTABLE2 VALUES (1, DATE '2015-07-01',0,2,120);
INSERT INTO MYTABLE2 VALUES (1, DATE '2015-08-01',0,2,120);
INSERT INTO MYTABLE2 VALUES (1, DATE '2015-09-01',0,2,120);
INSERT INTO MYTABLE2 VALUES (1, DATE '2015-10-01',0,2,120);
INSERT INTO MYTABLE2 VALUES (1, DATE '2015-11-01',240,12,240);
INSERT INTO MYTABLE2 VALUES (1, DATE '2016-12-01',240,12,240);      --newly added records
INSERT INTO MYTABLE2 VALUES (1, DATE '2016-01-01',960,4,240);      --newly added records
INSERT INTO MYTABLE2 VALUES (1, DATE '2016-02-01',0,4,240);      --newly added records
INSERT INTO MYTABLE2 VALUES (1, DATE '2016-03-01',0,4,240);      --newly added records
INSERT INTO MYTABLE2 VALUES (1, DATE '2016-04-01',0,4,240);      --newly added records
INSERT INTO MYTABLE2 VALUES (1, DATE '2016-05-01',960,4,240);      --newly added records
INSERT INTO MYTABLE2 VALUES (1, DATE '2016-06-01',0,4,240);      --newly added records
INSERT INTO MYTABLE2 VALUES (1, DATE '2016-07-01',0,4,240);      --newly added records
INSERT INTO MYTABLE2 VALUES (1, DATE '2016-08-01',0,4,240);      --newly added records
INSERT INTO MYTABLE2 VALUES (2, DATE '2015-01-01',60,3,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2015-02-01',0,3,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2015-03-01',0,3,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2015-04-01',0,3,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2015-05-01',180,1,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2015-06-01',0,1,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2015-07-01',0,1,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2015-08-01',0,1,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2015-09-01',0,1,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2015-10-01',0,1,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2015-11-01',0,1,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2015-12-01',0,1,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2016-01-01',0,1,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2016-02-01',0,1,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2016-03-01',0,1,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2016-04-01',0,1,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2016-05-01',15,12,15);
INSERT INTO MYTABLE2 VALUES (2, DATE '2016-06-01',15,12,15);
SELECT * FROM MYTABLE2;

Answer 1

我认为计算是：

select t.*,
       premium / (12 / yearly_payment_freq)) as monthly_premium_calculated
from mytable t;

编辑：

我明白了，您还需要在中间月份进行此分布。因此，您可以通过计算非零付款的数量来分配组。那么：

select t.*,
       ( max(premium) over (partition by policy_id, grp) /
         (12 / yearly_payment_freq)
       ) as monthly_premium_calculated
from (select t.*,
             sum(case when premium > 0 then 1 else 0 end) over (partition by policy_id order by payment_date) as grp
      from mytable t
     ) t;

Here 是一个 dbfiddle（它使用 Postgres，因为它比 Oracle 更容易设置）。