sql continuous column values from data for pivot table view - sql-server

Question

I have data in a table in this structure:
Region | Date | Value
A | 01/01/2014 | 100
A | 01/20/2014 | 50
A | 01/02/2014 | 200
A | 01/05/2014 | 300
B | 01/01/2014 | 50
B | 02/15/2014 | 70
B | 02/25/2014 | 50
C | 05/02/2014 | 70
I am trying to create a pivot view like this using T-SQL queries:
Region | Jan-2014 | Feb-2014 | Mar-014 | Apr-2014 | May-2014 | -> thru desired month-year
A | 150 | 200 | 0 | 0 | 300 |
B | 50 | 120 | 0 | 0 | 0 |
C | 0 | 0 | 0 | 0 | 70 |
Please note, multiple values in the same month for a given region needs to be aggregated
Months that have no records should still show up as columns with zero values (Ex: March and April)
I tired using pivot options, withroll up etc., -- but can't seem to get this to work
any help is much appreciated..
Thank you.

Answer 1

The reason why you do not have results for those months is because you are missing the dates that you are pivoting into columns.
There are a few ways that you can do this. You can hard code all of the date values in the IN portion of your query so the columns will appear:
select Region,
isnull([Jan-2014], 0) [Jan-2014], isnull([Feb-2014], 0) [Feb-2014],
isnull([Mar-2014], 0) [Mar-2014], isnull([Apr-2014], 0) [Apr-2014],
isnull([May-2014], 0) [May-2014], isnull([Jun-2014], 0) [Jun-2014],
isnull([Jul-2014], 0) [Jul-2014], isnull([Aug-2014], 0) [Aug-2014],
isnull([Sep-2014], 0) [Sep-2014], isnull([Oct-2014], 0) [Oct-2014],
isnull([Nov-2014], 0) [Nov-2014], isnull([Dec-2014], 0) [Dec-2014]
from
(
select left(datename(month, t.date), 3) +'-'
+ cast(year(t.date) as char(4)) monthYear,
t.region,
t.value
from yt t
) src
pivot
(
sum(value)
for monthYear in ([Jan-2014], [Feb-2014], [Mar-2014], [Apr-2014],
[May-2014], [Jun-2014], [Jul-2014], [Aug-2014],
[Sep-2014], [Oct-2014], [Nov-2014], [Dec-2014])
) piv;
See SQL Fiddle with Demo.
Another way to do this is to create a table of dates that you can use in your query. Once created, then you can use a LEFT JOIN to your table so you return all of the dates that you want to appear. You can either create a recursive query to generate this list in your PIVOT query, or populate a table with a list of dates. A recursive query will be similar to this:
;with dates (startDate, endDate) as
(
select min(date), cast('2014-12-31' as date)
from yt
union all
select dateadd(m, 1, startDate), enddate
from dates
where month(startDate) + 1 <= month(enddate)
)
select startDate
from dates;
See SQL Fiddle with Demo. If you join this to your PIVOT, then the query will be:
select *
from
(
select left(datename(month, d.startdate), 3) +'-'
+ cast(year(d.startdate) as char(4)) monthYear,
t.region,
t.value
from dates d
left join yt t
on month(d.startdate) = month(t.date)
and year(d.startdate) = year(t.date)
) src
pivot
(
sum(value)
for monthYear in ([Jan-2014], [Feb-2014], [Mar-2014], [Apr-2014],
[May-2014], [Jun-2014], [Jul-2014], [Aug-2014],
[Sep-2014], [Oct-2014], [Nov-2014], [Dec-2014])
) piv
where region is not null;
See SQL Fiddle with Demo. You could use this as a part of your PIVOT query, but you would still need to hand-code all of the dates in the IN clause for the PIVOT.
I am guessing that you want to use a dynamic SQL version of this query so the dates will change based on your needs. The dynamic version will be:
DECLARE #cols AS NVARCHAR(MAX),
#colsNull AS NVARCHAR(MAX),
#query AS NVARCHAR(MAX),
#startdate datetime = '2014-01-01',
#enddate datetime = '2014-12-01'
;with dates (startDate, endDate) as
(
select #startdate, #enddate
from yt
union all
select dateadd(m, 1, startDate), enddate
from dates
where dateadd(m, 1, startDate) <= enddate
)
select #cols = STUFF((SELECT ',' + QUOTENAME(left(datename(month, d.startdate), 3) +'-'
+ cast(year(d.startdate) as char(4)))
from dates d
-- where startdate >= '2014-01-01' and startdate <= '2014-06-01'
group by d.startdate
order by d.startdate
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)')
,1,1,''),
#colsNull = STUFF((SELECT ', isnull(' + QUOTENAME(left(datename(month, d.startdate), 3) +'-'
+ cast(year(d.startdate) as char(4)))+', 0) as '+QUOTENAME(left(datename(month, d.startdate), 3) +'-'
+ cast(year(d.startdate) as char(4)))
from dates d
-- where startdate >= '2014-01-01' and startdate <= '2014-06-01'
group by d.startdate
order by d.startdate
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)')
,1,1,'')
from dates
set #query = 'SELECT region, ' + #colsNull + ' from
(
select left(datename(month, t.date), 3) +''-''
+ cast(year(t.date) as char(4)) monthYear,
t.region,
t.value
from yt t
) x
pivot
(
sum(value)
for monthyear in (' + #cols + ')
) p '
execute(#query);
See SQL Fiddle with Demo. This version uses the recursive CTE to generate the list of dates that will be used in the dynamic sql string. Even though your table of data might not exist for the months displayed, you will still have a new column.
This gives a result:
| REGION | JAN-2014 | FEB-2014 | MAR-2014 | APR-2014 | MAY-2014 | JUN-2014 | JUL-2014 | AUG-2014 | SEP-2014 | OCT-2014 | NOV-2014 | DEC-2014 |
----------------------------------------------------------------------------------------------------------------------------------------------
| A | 650 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| B | 50 | 120 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| C | 0 | 0 | 0 | 0 | 70 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |