SQL pivot not handling absence of data - sql-server

Question

I have a table
Create table dbo.temp(ID int, ExpenseID int, ExpenseName Varchar(100) null, ExpenseDate date null, Value decimal(19,4) null)
Insert into dbo.temp values(1,100,'CostClothes','2015-01-01',100.0000)
Insert into dbo.temp values(1,200,'Discount','2015-01-01',1.0000)
Insert into dbo.temp values(2,100,'CostClothes','2016-01-01',250.0000)
Insert into dbo.temp values(2,200,'Discount','2016-01-01',1.0000)
Insert into dbo.temp values(1,100,'CostClothes','2014-01-01',500.0000)
Insert into dbo.temp values(2,200,'Discount','2014-01-01',5.0000)
Now I want to pivot this table on ExpenseID and the expected output is
Create table dbo.output(ID int, CostClothes decimal(19,4) null, Discount decimal(19,4) null, ExpenseDate date null)
insert into dbo.output values (1,100.0000,1.0000,'2015-01-01')
insert into dbo.output values (1,500.0000,NULL,'2014-01-01')
insert into dbo.output values (2,NULL,5.0000,'2014-01-01')
insert into dbo.output values (2,100.0000,1.0000,'2016-01-01')
This is what I have and i am not getting correct output
SELECT ID,ISNULL([100],0) as CostClothes,ISNULL([200],0) as Discount,expenseDate
FROM
(
SELECT * FROM dbo.temp
) AS p
PIVOT
(
MAX(Value) for ExpenseID in ([100],[200])
) AS PV1
How do I change the query
Thanks
MR

Answer 1

Given the sample data, you can either do a pivot with an aggregation, or just a simple conditional aggregation.
select
ID
,max(case when ExpenseName = 'CostClothes' then Value end) as CostClothes
,max(case when ExpenseName = 'Discount' then Value end) as Discount
,ExpenseDate
from
dbo.temp
group by
ID
,ExpenseDate
order by
ID
Pivot Method
select
ID
,CostClothes = max(CostClothes)
,Discount = max(Discount)
,ExpenseDate
from
dbo.temp
pivot(
max(Value) for ExpenseName in (CostClothes, Discount)
) p
group by
ID
,ExpenseDate
order by
ID
Demo: http://rextester.com/ENHTIK13664

Answer 2

Your problem is that you're selecting too much information in your derived table.
If you're going to use Select *, you don't even need a derived table. When using a PIVOT, you should select only the fields you need in your final result. The same way you would write the conditional aggregation query.
SELECT ID,
ISNULL([100], 0) AS CostClothes,
ISNULL([200], 0) AS Discount,
expenseDate
FROM
(
SELECT ID,
ExpenseID,
ExpenseDate,
Value
FROM temp
) AS p PIVOT(MAX(Value) FOR ExpenseID IN([100],[200])) AS PV1
This should give you the result you want by eliminating the ExpenseName column. Or you could use the ExpenseName column and eliminate the ExpenseID column