用于查看所有先前行的 SQL 语句

Posted 2023-03-31

【中文标题】用于查看所有先前行的 SQL 语句

【英文标题】：SQL statement to look at ALL the previous rows

【发布时间】：2017-01-17 18:41:21

【问题描述】：

我有下表

create table interest_earning_assets (
  key integer
  month_of varchar(7),
  principal_distribution numeric(38,2),
  closed_loan_amount numeric(38,2)
);

数据看起来像这样

key month_of    principal_distribution  closed_loan_amount
24  2017-01     4133500.00              5984695.00
23  2016-12     12018303.93             26941275.40
22  2016-11     6043945.46              21239620.20
21  2016-10     2864195.39              20368518.20

我有两个要求。

总结closed_amount_values

对于每个月（当前为 24 个月，下个月为 25 个月，然后是 26 个月等），我需要将 closed_amount_values 与前几个月 all 的值相加，即

2017-01 sum(closed_loan_amount for 2016-10 + 2016-11 + 2016-12 + 2017-01)
2016-12 sum(closed_loan_amount for 2016-10 + 2016-11 + 2016-12)
2016-11 sum(closed_loan_amount for 2016-10 + 2016-11) 
2016-10 sum(closed_loan_amount for 2016-10 )  

将closed_loan_amount 的总和减去principal_distribution

一旦我得到了汇总值，我需要将每个月的principal_distribution 减去closed_loan_amount 的总和

2017-01 principal_distribution for 2017-01 - sum(closed_loan_amount for 2016-10 + 2016-11 + 2016-12 + 2017-01)
2016-12 principal_distribution for 2016-12 - sum(closed_loan_amount for 2016-10 + 2016-11 + 2016-12)
2016-11 principal_distribution for 2016-11 - sum(closed_loan_amount for 2016-10 + 2016-11) 
2016-10 principal_distribution for 2016-10 - sum(closed_loan_amount for 2016-10 )  

Redshift 不支持存储过程，我不精通 Python。所以我试图这样使用滞后

select month_of, closed_loan_amount, 
       lag(closed_loan_amount,1) over (order by month_of desc) as previous_month 
from public.interest_earning_assets

它有效，但只给我上个月的值。我也在考虑使用 CTE，但我今天刚接到这个任务。我怎样才能在 SQL 中做到这一点？

【问题讨论】：

sum(closed_loan_amount) over (order by month-of rows between unbounded preceding and current row) 我不知道 Redshift 支持这些操作的程度。 docs.aws.amazon.com/redshift/latest/dg/…

【参考方案1】：

使用带有窗口规范的sum 窗口函数来查看所有前面的行以获得 closed_loan_amount 的总和并从 principal_distribution 中减去它。

select month_of, closed_loan_amount, 
 principal_distribution
-sum(closed_loan_amount) over (order by month_of desc rows between current row and unbounded following) as some_value
from public.interest_earning_assets

【讨论】：

我需要学习像这样的窗口函数。我不知道我能做到这一点。谢谢你。接受并投票赞成。谢谢！！！

【参考方案2】：

试试这个：

SELECT [key], month_of,
       SUM(closed_loan_amount) OVER(ORDER BY month_of),
       principal_distribution + SUM(closed_loan_amount) OVER(ORDER BY month_of)
FROM interest_earning_assets

带有ORDER BY 子句的SUM 的窗口版本根据ORDER BY 子句中出现的第二个字段定义的顺序计算字段的运行总计。