SQL Join with the Same table with out Pivot Table - sql-server

Question

I have a problem, I'm trying to join the same table and make column of it. For example the table contains id, name, type.
And the data can be:
id name type Date
--------------------------------------------------
1 KKKK BP 05/05/2017
2 MMMM KS 07/10/2016
3 LLL TL 04/05/2017
4 NNN BP 06/01/2016
I want to make a table with the following design:
- id name BP KS TL
-------------------------------------------------------------
1 KKK 05/05/2017
2 MMM 07/10/2016
3 LLL 04/05/2017
4 NNN 06/01/2017
I tried Pivot table and didn't work.
Any idea?

Answer 1

You can use the PIVOT function in SQL Server 2008 to convert your rows of data into columns:
select
id,
name,
BP, KS, TL
from
(
select id,
name,
type,
[date]
from mytable
) d
pivot
(
max([date])
for type in (BP, KS, TL)
) piv;
This could also be written using conditional logic, like a CASE expression with some aggregation:
select id,
name,
BP = max(case when type = 'BP' then [date] end),
KS = max(case when type = 'KS' then [date] end),
TL = max(case when type = 'TL' then [date] end)
from mytable
group by id, name;
Based on your comment that you could have multiple dates for each name and type combination, you can still use similar queries you'll just have to use a windowing function like row_number to get the final result you want.
If you want to use the conditional logic version, you'd change the query to be:
select
name,
BP1 = max(case when type = 'BP' and rn =1 then [date] end),
BP2 = max(case when type = 'BP' and rn =2 then [date] end),
BP3 = max(case when type = 'BP' and rn =3 then [date] end),
KS1 = max(case when type = 'KS' and rn =1 then [date] end),
KS2 = max(case when type = 'KS' and rn =2 then [date] end),
TL1 = max(case when type = 'TL' and rn =1 then [date] end)
from
(
select
name,
[type],
[date],
rn = row_number() over(partition by name, [type] order by [date] desc)
from mytable
) d
group by name;
The PIVOT version would be:
select
name,
BP1, BP2, BP3, KS1, KS2, TL1
from
(
select
name,
type = type + cast(rn as varchar(2)),
[date]
from
(
select
name,
type,
[date],
rn = row_number() over(partition by name, [type] order by [date] desc)
from mytable
)s
) d
pivot
(
max([date])
for type in (BP1, BP2, BP3, KS1, KS2, TL1)
) piv;
As you can see there is a lot of typing to get all of these columns, so you could use dynamic SQL to get the final result:
DECLARE #cols AS NVARCHAR(MAX),
#query AS NVARCHAR(MAX)
select #cols = STUFF((SELECT ',' + QUOTENAME(Type + cast(rn as varchar(2)))
from
(
select type,
rn = row_number() over(partition by name, type order by date desc)
from mytable
) d
group by type, rn
order by type, rn
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)')
,1,1,'')
set #query = N'SELECT name, ' + #cols + N'
from
(
select
name,
type = type + cast(rn as varchar(2)),
[date]
from
(
select
name,
type,
[date],
rn = row_number() over(partition by name, [type] order by [date] desc)
from mytable
)s
) x
pivot
(
max(date)
for type in (' + #cols + N')
) p '
exec sp_executesql #query;
I created a demo to show that they all return the same result.