How to pick the 2 highest prices paid - sql

Question

This is the input data.
Dept Company Code Payment Amt
Gardeners Sort:Insurance Carrier 100 20.00
Gardeners Sort:Insurance Carrier 100 22.00
Gardeners Sort:Insurance Carrier 100 21.00
Gardeners Sort:Insurance Carrier 100 20.00
Gardeners Sort:Insurance Carrier 100 22.00
I want to return
Sort:Insurance Carrier 100 - 22.00 and 21.00
Not 22.00 and 22.00 I fear this code is returning 22 and 22 arguably the 2 top prices paid but not really so.
I have this SQL
SELECT
[DEPT], [Sort: Procedure Code] as Code, [Sort: Insurance Carrier],
SUM(CASE WHEN num = 1 THEN [Pmt Amount] ELSE 0 END) AS [first high],
SUM(CASE WHEN num = 2 THEN [Pmt Amount] ELSE 0 END) AS [second high]
FROM
(
SELECT ROW_NUMBER() OVER(PARTITION BY
[DEPT], [Sort: Procedure Code], [Sort: Insurance Carrier]
ORDER BY [Pmt Amount] DESC) AS num,
[DEPT], [Sort: Procedure Code], [Sort: Insurance Carrier],
[Pmt Amount]
FROM
[revenuedetail$]
) AS t
WHERE num IN (1, 2)
GROUP BY [DEPT], [Sort: Procedure Code], [Sort: Insurance Carrier]

Answer 1

If you want the same value to have the same number, then you should use dense_rank() instead of row_number(). But you are on the right track!
Also change sum() to max() to avoid summing the values with the same dense_rank().
Try this:
select
[dept]
, [Sort: Procedure Code] as Code
, [Sort: Insurance Carrier]
, max(case when num = 1 then [Pmt Amount] else 0 end) as [first high]
, max(case when num = 2 then [Pmt Amount] else 0 end) as [second high]
from (
select
dense_rank() over(
partition by [dept], [Sort: Procedure Code], [Sort: Insurance Carrier]
order by [Pmt Amount] desc
) as num
, [dept]
, [Sort: Procedure Code]
, [Sort: Insurance Carrier]
, [Pmt Amount]
from [revenuedetail$]
) as t
where num in (1, 2)
group by [dept], [Sort: Procedure Code], [Sort: Insurance Carrier]
rextester demo: http://rextester.com/PJCDDC90476
returns:
+-----------+------+-------------------------+------------+-------------+
| dept | Code | Sort: Insurance Carrier | first high | second high |
+-----------+------+-------------------------+------------+-------------+
| Gardeners | 100 | Sort:Insurance Carrier | 22.00 | 21.00 |
+-----------+------+-------------------------+------------+-------------+

Answer 2

You seem to want dense_rank() rather than row_number():
SELECT [DEPT], [Sort: Procedure Code] as Code, [Sort: Insurance Carrier],
SUM(CASE WHEN num = 1 THEN [Pmt Amount] END) AS [first high],
SUM(CASE WHEN num = 2 THEN [Pmt Amount] END) AS [second high]
FROM (SELECT DENSE_RANK() OVER (PARTITION BY [DEPT], [Sort: Procedure Code], [Sort: Insurance Carrier]
ORDER BY [Pmt Amount] DESC
) AS num,
rd.*
FROM [revenuedetail$] rd
) rd
WHERE num IN (1, 2)
GROUP BY [DEPT], [Sort: Procedure Code], [Sort: Insurance Carrier];
Notes:
I removed the ELSE 0. If there is no second value, then this version returns NULL rather than 0. I find that more intuitive (add the ELSE 0 back if that is not the behavior you want).
I added more meaningful table aliases. rd makes more sense than t.
I used rd.* in the subquery. That actually shortens the query and makes it easier to modify.
You should reconsider your column names. All the square braces just make the code harder to write and to read.

Answer 3

If the sql server version is 2012 and above, Then Lead() can be used:
select Top 1 [DEPT], [Sort: Procedure Code], [Sort: Insurance Carrier],
[Pmt Amount] AS [first high],
Lead([Pmt Amount],1)over(partition by [DEPT], [Sort: Procedure Code],
[Sort: Insurance Carrier] ORDER BY [Pmt Amount] DESC)AS [Second high]
from [revenuedetail$] order by [Pmt Amount] desc