Count and agregate data by time - sql

Question

I'm looking for a beauty-easy to read-smart SQL query (SQLite engine) to agregate data in column. It is more easy to explain with an example :
Data table :
id elapsedtime httpcode
1 0.0 200
2 0.1 200
3 0.3 301
4 0.6 404
5 1.0 200
6 1.1 404
7 1.2 500
Expected result set : a column by httpcode, with number of code by time. In this example, the time agregation is 0.2s (but it can be agregated at a second, or 10s). I'm interested only in some expected http_code :
time code_200 code_404 code_500 code_other
0.0 2 0 0 0
0.2 0 0 0 1
0.4 0 1 0 0
0.6 0 0 0 0
0.8 0 0 0 0
1.0 1 1 1 0
It is not mandatory for "time" to be continuous. In the previous example, line with time 0.6 and 0.6 can be removed.
For the moment, I can do this by doing 4 different requests (one by http code) and I agregate the result in my developped application:
select
0.2 * cast (elapsedtime/ 0.2 as int) as time, count(id) as code_200
from test
where (httpcode=200)
group by time
But i'm pretty sure i can achieve this with a single query. Unfortunally i'm not mastering UNION keyword...
Is there a way to get such data in a single SELECT ?
See SQLFiddle : http://sqlfiddle.com/#!5/2081f/3/1

Answer 1

Found a nicer solution than my original post, which I'll leave in, in case you're curious. Here's the nicer solution:
with t1 as (
select
0.2 * cast (elapsedtime/ 0.2 as int) as time,
case httpcode when 200 then 1 else 0 end code_200,
case httpcode when 404 then 1 else 0 end code_404,
case httpcode when 500 then 1 else 0 end code_500,
case when httpcode not in (200, 404, 500) then 1 else 0 end code_other
from test
)
select time,
sum(code_200) as count_200,
sum(code_404) as count_404,
sum(code_500) as count_500,
sum(code_other) as count_other
from t1
group by time;
Old solution:
This might not be too easy on the eye, but it more or less works (only difference between your desired output and what I get with this is that time groupings that have no values (0.6 and 0.8 in your example) are omitted:
with
t_all as (select
0.2 * cast (elapsedtime/ 0.2 as int) as time, count(id) as total
from test
group by time
),
t_200 as (select
0.2 * cast (elapsedtime/ 0.2 as int) as time, count(id) as code_200
from test
where (httpcode=200)
group by time),
t_404 as (select
0.2 * cast (elapsedtime/ 0.2 as int) as time, count(id) as code_404
from test
where (httpcode=404)
group by time),
t_500 as (select
0.2 * cast (elapsedtime/ 0.2 as int) as time, count(id) as code_500
from test
where (httpcode=500)
group by time),
t_other as (select
0.2 * cast (elapsedtime/ 0.2 as int) as time, count(id) as code_other
from test
where (httpcode not in (200, 404, 500))
group by time)
select
t_all.time,
total,
ifnull(code_200,0) as count_200,
ifnull(code_404,0) as count_404,
ifnull(code_500,0) as count_500,
ifnull(code_other,0) as count_other
from t_all
left join t_200 on t_all.time = t_200.time
left join t_404 on t_all.time = t_404.time
left join t_500 on t_all.time = t_500.time
left join t_other on t_all.time = t_other.time;

Answer 2

It may help you
select
0.2 * cast (elapsedtime/ 0.2 as int) as time, count(id) as code_200,
0 as code_404,
0 as code_500,
0 as code_other
from
test
where (httpcode=200)
group by time
union
select
0.2 * cast (elapsedtime/ 0.2 as int) as time,0 as code_200,
count(id) as code_404,
0 as code_500,
0 as code_other
from
test
where (httpcode=404)
group by time
union
select
0.2 * cast (elapsedtime/ 0.2 as int) as time,0 as code_200,
0 as code_400,
count(id) as code_500,
0 as code_other
from
test
where (httpcode=500)
group by time
union
select
0.2 * cast (elapsedtime/ 0.2 as int) as time,0 as code_200,
0 as code_400,
0 as code_500,
count(id) as code_other
from
test
where (httpcode<>200 and httpcode <> 404 and httpcode <> 500)
group by time