Reputation: 2570
Very simple task but I fail to write the corresponding SQL.
Two factor variables, var1 with 5 levels, var2 is binary with 0/1.
I want to count all rows grouped by var1 and var2, but with new variables.
select
var1, var2, count(*)
from
tab
group by
var1, var2;
gives me
var1 var2 Count(*)
1 1 0 32
2 1 1 80
3 2 1 80
4 2 0 33
5 3 1 82
6 3 0 33
7 4 1 81
8 4 0 33
9 5 0 33
10 5 1 88
Desired is this format:
var1 var1=0 var1=1
1 32 80
2 33 80
3 33 82
4 33 81
5 33 88
Does not work:
select
var1,
case
when var2 = 1 then count(*)
end as svar1,
case
when var2 = 0 then count(*)
end as svar2
from
tab
group by
var1;
My query does not work, because var2 is not part of the associated group, how do I get this format?
Upvotes: 1
Views: 2831
Reputation: 31993
you can try by using this query
select var1,0 as [var1=0],1 as [var1=1]
from(
select var1,var2,QTY from
(
select
var1,
var2,
count(*) as QTY
from tab
group by var1,var2;
) )up pivot (sum(QTY) for var1 in(0,1)) as pvt
order by var1
Upvotes: 0
Reputation: 28900
You need to do coniditional aggregation..please provide some test data to test as well
select var1
sum(case when var1=0 then 1 else 0 end) as var1,
sum(case when var1=1 then 1 else 0 end) as var2
group by var1
Upvotes: 3
Reputation: 538
try this
select
var1,
case
when var2 = 1 then count(*)
end as svar1,
case
when var2 = 0 then count(*)
end as svar2
from tab
group by var1,var2;
Upvotes: -1