unduplicating similar data based on multiple fields - sql

Question

I'd like to create a table of possible duplicate records from an original table, however the data is based on two different attibutes and they must only join in the same grouping id. Also in some cases the data doesn't look exactly the same (but the similarities are there). Here is what the original table would look like:
group_id| House_num | Apt | code
----------------------------------
45 | 1000 | 1 | M
45 | 1 | | D
45 | 1000 | 2 | M
45 | 2 | | D
87 | 2300 | 310 | M
87 | 2310 | | D
87 | 2400 | 470 | M
87 | 2470 | | D
What I'd like to be returned is where these like numbers are all on the same row. So something like:
new_id |group_id|a.house_num|a.apt|a.code|b.house_num|b.apt| b.code
-------------------------------------------------------------------------
1 | 45 | 1000 | 1 | M | 1 | | D
2 | 45 | 1000 | 2 | M | 2 | | D
3 | 87 | 2300 | 310 | M | 2310 | | D
4 | 87 | 2400 | 470 | M | 2470 | | D
I'm not sure what kind of join to use here; also, im not sure how to get around the ones where a.house_num is the base number, a.apt is the suffixed number and b.housenumber is the combination of the two. Any help would be greatly appreciated, thank you.

Answer 1

create table t (group_id, House_num , Apt , code) as
select 45 , 1000 , 1 , 'M' from dual union all
select 45 , 1 , null , 'D' from dual union all
select 45 , 1000 , 2 , 'M' from dual union all
select 45 , 2 , null , 'D' from dual union all
select 87 , 2300 , 310 , 'M' from dual union all
select 87 , 2310 , null , 'D' from dual union all
select 87 , 2400 , 470 , 'M' from dual union all
select 87 , 2470 , null , 'D' from dual;
select rownum new_id,
a.GROUP_ID, a.HOUSE_NUM ahn, a.APT aapt, a.CODE acode,
b.HOUSE_NUM bhn, b.APT bapt, b.CODE bcode
from t a
join t b
on a.group_id = b.group_id
and a.code = 'M'
and b.code = 'D'
and (
b.house_num = a.apt
or b.house_num like '%'||a.apt
);
NEW_ID GROUP_ID AHN AAPT ACODE BHN BAPT BCODE
1 45 1000 1 M 1 D
2 45 1000 2 M 2 D
3 87 2300 310 M 2310 D
4 87 2400 470 M 2470 D

Answer 2

You can use the below to achieve your answer.The DB fiddle here
WITH data
AS (SELECT * FROM YOUR_TABLE),
data2
AS (SELECT row_number()over
(order by thing1,thing2) rw,
d1.group_id,
d1.thing1 a_thing1,
d1.thing2 a_thing2,
d1.thing3 a_thing3
FROM data d1
WHERE d1.thing3 = 'M'),
data3
AS (SELECT row_number()over
(order by thing1,thing2) rw,
d1.group_id,
d1.thing1 b_thing1,
d1.thing2 b_thing2,
d1.thing3 b_thing3
FROM data d1
WHERE d1.thing3 = 'D')
SELECT d2.rw New_id,
d2.group_id,
d2.a_thing1,
d2.a_thing2,
d2.a_thing3,
d3.b_thing1,
d3.b_thing2,
d3.b_thing3
FROM data2 d2,
data3 d3
WHERE d2.group_id = d3.group_id
AND d2.rw = d3.rw