SUMIFS (Sum if with multiple conditions) with SQL Server - sql

Question

This should be done easy in Excel, however, I would like to have this kind of calculation done via SQL. I could use the GROUP BY , OVER() to calculate the SUM and % of a single year. But I failed to present the data 3 years at once. Any help will be appreciated.

Answer 1

Since you are using SQL Server, if you are using SQL Server 2005+ then you can use the PIVOT function to get the result. This solution implements both an unpivot and a pivot process to get the result. The starting point for this result is to calculate the total percent and total by type:
select type, year, total,
round(total / sum(total) over(partition by year)*100.0, 1) t_per,
sum(total) over(partition by type) t_type,
round(sum(total) over(partition by type)*100.0/sum(total) over(), 1) tot_per
from tablea
See SQL Fiddle with Demo. This will give a result with multiple columns that you want to pivot so you can unpivot the data into multiple rows using CROSS APPLY:
select type,
col = cast(year as varchar(4))+'_'+col,
value,
t_type
from
(
select type, year, total,
round(total / sum(total) over(partition by year)*100.0, 1) t_per,
sum(total) over(partition by type) t_type,
round(sum(total) over(partition by type)*100.0/sum(total) over(), 1) tot_per
from tablea
) d
cross apply
(
select 'total', total union all
select 't_per', t_per
) c (col, value);
See Demo. Finally you can apply the PIVOT function to the values in col:
select type,
[2010_total], [2010_t_per],
[2011_total], [2011_t_per],
[2012_total], [2012_t_per],
t_type,
tot_per
from
(
select type,
col = cast(year as varchar(4))+'_'+col,
value,
t_type,
tot_per
from
(
select type, year, total,
round(total / sum(total) over(partition by year)*100.0, 1) t_per,
sum(total) over(partition by type) t_type,
round(sum(total) over(partition by type)*100.0/sum(total) over(), 1) tot_per
from tablea
) d
cross apply
(
select 'total', total union all
select 't_per', t_per
) c (col, value)
) s
pivot
(
max(value)
for col in ([2010_total], [2010_t_per],
[2011_total], [2011_t_per],
[2012_total], [2012_t_per])
) piv
See SQL Fiddle with Demo. This could be refactored to use a CTE instead of the subqueries and this could also be converted to use dynamic SQL if the year will be unknown.
If you have an unknown number of values, then the dynamic SQL code will be:
DECLARE #cols AS NVARCHAR(MAX),
#query AS NVARCHAR(MAX)
select #cols = STUFF((SELECT ',' + QUOTENAME(cast(year as varchar(4))+'_'+col)
from tablea
cross apply
(
select 'total', 1 union all
select 't_per', 2
) c (col, so)
group by year, col, so
order by year, so
FOR XML PATH(''), TYPE
).value('.', 'NVARCHAR(MAX)')
,1,1,'')
set #query = 'SELECT type,' + #cols + ', t_type, tot_per
from
(
select type,
col = cast(year as varchar(4))+''_''+col,
value,
t_type,
tot_per
from
(
select type, year, total,
round(total / sum(total) over(partition by year)*100.0, 1) t_per,
sum(total) over(partition by type) t_type,
round(sum(total) over(partition by type)*100.0/sum(total) over(), 1) tot_per
from tablea
) d
cross apply
(
select ''total'', total union all
select ''t_per'', t_per
) c (col, value)
) x
pivot
(
max(value)
for col in (' + #cols + ')
) p '
execute sp_executesql #query;
See Demo. Both the static version and the dynamic version give the result:
| TYPE | 2010_TOTAL | 2010_T_PER | 2011_TOTAL | 2011_T_PER | 2012_TOTAL | 2012_T_PER | T_TYPE | TOT_PER |
---------------------------------------------------------------------------------------------------------
| A | 1 | 16.7 | 1 | 16.7 | 1 | 16.7 | 3 | 16.7 |
| B | 2 | 33.3 | 2 | 33.3 | 2 | 33.3 | 6 | 33.3 |
| C | 3 | 50 | 3 | 50 | 3 | 50 | 9 | 50 |

Answer 2

SUMIF can be replicated in SQL with SUM(case statement):
SELECT Type
,SUM(CASE WHEN Year = '2010' THEN Total ELSE 0 END)'2010 Total'
,SUM(CASE WHEN Year = '2010' THEN Total ELSE 0 END)*1.0/SUM(SUM(CASE WHEN Year = '2010' THEN Total ELSE 0 END)) OVER () '2010 Percent of Total'
,SUM(CASE WHEN Year = '2011' THEN Total ELSE 0 END)'2011 Total'
,SUM(CASE WHEN Year = '2011' THEN Total ELSE 0 END)*1.0/SUM(SUM(CASE WHEN Year = '2011' THEN Total ELSE 0 END)) OVER () '2011 Percent of Total'
,SUM(CASE WHEN Year = '2012' THEN Total ELSE 0 END)'2012 Total'
,SUM(CASE WHEN Year = '2012' THEN Total ELSE 0 END)*1.0/SUM(SUM(CASE WHEN Year = '2012' THEN Total ELSE 0 END)) OVER () '2012 Percent of Total'
,SUM(Total) 'Total'
,SUM(Total)*1.0/SUM(SUM(Total)) OVER () 'Percent of Total'
FROM Table
GROUP BY Type

Answer 3

For the sake of simplicity I would prefer to show this result vertically:
SELECT
Type,
Year,
SUM(Total) as Dollars,
ROUND(SUM(Total) * 100 / (SELECT SUM(TOTAL) FROM TableA t2 WHERE t2.Year = t1.Year),1) as Per
FROM TableA t1
Group By Type, Year
Output:
TYPE YEAR DOLLARS PERCENT
A 2010 1 16.7
B 2010 2 33.3
C 2010 3 50
A 2011 1 16.7
B 2011 2 33.3
C 2011 3 50
A 2012 1 16.7
B 2012 2 33.3
C 2012 3 50
Sql Fiddle Demo