问题描述
我有以下信息：
| 产品 | IS_BUY | 日期 |
|---|---|---|
| A | 1 | 01/01/2021 |
| A | 1 | 02/01/2021 |
| A | 1 | 03/01/2021 |
| A | 0 | 04/01/2021 |
| A | 1 | 05/01/2021 |
| A | 1 | 06/01/2021 |
| B | 1 | 01/01/2021 |
| C | 1 | 01/01/2021 |
| C | 0 | 02/01/2021 |
| C | 0 | 03/01/2021 |
| C | 1 | 04/01/2021 |
我需要在每个产品存在或不购买的日期之间留下以下结果：
| 产品 | IS_BUY | FROM_DATE | TO_DATE |
|---|---|---|---|
| A | 1 | 01/01/2021 | 04/01/2021 |
| A | 0 | 04/01/2021 | 05/01/2021 |
| A | 1 | 05/01/2021 | null |
| B | 1 | 01/01/2021 | null |
| C | 1 | 01/01/2021 | 02/01/2021 |
| C | 0 | 02/01/2021 | 04/01/2021 |
| C | 1 | 04/01/2021 | null |
感谢您的帮助！
解决方法
您没有告诉我们您的 Oracle 版本。下面我将展示两种解决方案：第一种是更高级的，使用 match_recognize 子句，在 Oracle 12.1 中添加，然后是较旧的方法，使用 Tabibitosan 方法解决间隙和岛屿问题（您的类问题属于）。
数据设置：
create table my_data (product,is_buy,eff_date) as
select 'A',1,to_date('01/01/2021','mm/dd/yyyy') from dual union all
select 'A',to_date('02/01/2021',to_date('03/01/2021',to_date('04/01/2021',to_date('05/01/2021',to_date('06/01/2021','mm/dd/yyyy') from dual union all
select 'B','mm/dd/yyyy') from dual union all
select 'C','mm/dd/yyyy') from dual
;
（顺便说一句，这是在帖子中包含示例数据的首选方式！）
注意 date 是保留关键字；我将列名称更改为 eff_date。
第一个解决方案，使用 match_recognize 匹配数据中的模式：
select product,eff_date,end_date
from my_data
match_recognize(
partition by product
order by eff_date
measures a.is_buy as is_buy,a.eff_date as eff_date,next(eff_date) as end_date
pattern ( a b* )
define b as is_buy = a.is_buy
);
PRODUCT IS_BUY EFF_DATE END_DATE
------- ------ ---------- ----------
A 1 01/01/2021 04/01/2021
A 0 04/01/2021 05/01/2021
A 1 05/01/2021
B 1 01/01/2021
C 1 01/01/2021 02/01/2021
C 0 02/01/2021 04/01/2021
C 1 04/01/2021
第二种解决方案，仅使用解析函数和聚合（Tabibitosan 方法）：
with prep as (
select product,lead(eff_date) over (partition by product
order by eff_date) as next_date,row_number() over (partition by product order by eff_date) -
row_number() over (partition by product,is_buy order by eff_date)
as grp
from my_data
)
select product,min(eff_date) as eff_date,max(next_date) keep (dense_rank last order by eff_date) as end_date
from prep
group by product,grp
order by product,eff_date
;
试试这个
Select
Product,Is_Buy,(select min(x.DATE) from products x where x.product=a.product and x.is_buy=a.is_buy) from_date,(select max(x.DATE) from products x where x.product=a.product and x.is_buy=a.is_buy) to_date,from products a
group by a.Product,a.Is_Buy