问题描述
此图像将向您显示问题结构
这应该是所需的输出
我想从重复记录中填充 Min Start_Date 和 Max End_Date (基于 Present 和 Absent 列)并制作一个主记录,与 Present 和 Absent 列的其他组合相同
我已经按照ID和开始日期进行了排序,以了解数据的行为。 我仅针对一个ID提到过这种情况,存在的其他ID也应提供相同类型的输出,如果我得到此示例的有效解决方案,则可以在整个表上实现正确的逻辑。 我尝试使用窗口功能,但未实现任何解决方案。 预先感谢
请注意,根据存在和不存在的列, ID 的重复记录数是可变的。
解决方法
这是一个空白问题。考虑使用行号之间的差异来构建“相邻”记录组,然后可以合并:
select id,min(start_date) start_date,max(end_date end_date,present,absent
from (
select t.*,row_number() over(partition by id order by start_date) rn1,row_number() over(partition by id,absent order by start_date) rn2
from mytable t
) t
group by id,absent,rn1 - rn2
order by 1,2
,
您可以使用MATCH_RECOGNIZE
:
SELECT *
FROM table_name
MATCH_RECOGNIZE (
PARTITION BY id
ORDER BY start_date
MEASURES FIRST( start_date ) AS start_date,MAX( end_date ) AS end_date,FIRST( present ) AS present,FIRST( absent ) AS absent
ONE ROW PER MATCH
PATTERN (FIRST_ROW EQUAL_ROWS*)
DEFINE EQUAL_ROWS AS
(
(
(
EQUAL_ROWS.present = PREV(EQUAL_ROWS.present)
) OR (
EQUAL_ROWS.present IS NULL AND PREV(EQUAL_ROWS.present) IS NULL
)
) AND (
(
EQUAL_ROWS.absent = PREV(EQUAL_ROWS.absent)
) OR (
EQUAL_ROWS.absent IS NULL AND PREV(EQUAL_ROWS.absent) IS NULL
)
)
)
)
因此,对于您的示例数据:
CREATE TABLE table_name ( id,start_date,end_date,absent ) AS
SELECT 1,DATE '2020-02-01',DATE '2020-03-01','Y','N' FROM DUAL UNION ALL
SELECT 1,DATE '2020-03-04',DATE '2020-04-19',DATE '2020-03-06',DATE '2020-03-09','N',DATE '2020-05-04',DATE '2020-09-04','Y' FROM DUAL UNION ALL
SELECT 1,DATE '2020-05-06',DATE '2020-06-26',DATE '2020-07-12',DATE '2020-08-12',NULL,NULL FROM DUAL UNION ALL
SELECT 1,DATE '2020-08-13',NULL FROM DUAL;
这将输出:
ID | START_DATE | END_DATE | PRESENT | ABSENT -: | :--------- | :-------- | :------ | :----- 1 | 01-FEB-20 | 19-APR-20 | Y | N 1 | 06-MAR-20 | 09-MAR-20 | N | N 1 | 04-MAY-20 | 04-SEP-20 | N | Y 1 | 12-JUL-20 | 12-AUG-20 | null | null
db 提琴here