Reputation: 35
I have a table named 'ward_councillors' and two columns 'ward' and 'councillor'. This table contains a list of all of the councillors and the ward they are responsible for. Each ward has three councillors and what i'd need to do is group by so that I have one distinct record for each ward and three new columns with the three councillors responsible for that ward in each.
For example currently have:
WARD | COUNCILLOR
ward a | cllr 1
ward a | cllr 2
ward a | cllr 23
ward b | cllr 4
ward b | cllr 5
ward b | cllr 8
trying to achieve:
WARD | COUNCILLOR 1| COUNCILLOR 2| COUNCILLOR 3
ward a | cllr 1 | cllr 2 | cllr 23
ward b | cllr 4 | cllr 5 | cllr 8
Thanks in advance, Chris
Upvotes: 2
Views: 118
Reputation: 3983
I have a problem with sqlFiddle, but I wrote this on my local postgres:
CREATE TABLE ward_councillors (
ward VARCHAR
, councillor VARCHAR
);
INSERT INTO ward_councillors
VALUES
('ward a', 'cllr 1')
, ('ward a', 'cllr 2')
, ('ward a', 'cllr 23')
, ('ward b', 'cllr 4')
, ('ward b', 'cllr 5')
, ('ward b', 'cllr 8')
;
SELECT *
FROM crosstab('SELECT ward, null /*doesnt really matter*/, councillor FROM ward_councillors')
AS (ward VARCHAR, councillor_1 VARCHAR, councillor_2 VARCHAR, councillor_3 VARCHAR, councillor_4 VARCHAR)
;
Seems to output just what you wanted:
ward | councillor_1 | councillor_2 | councillor_3 | councillor_4
--------+--------------+--------------+--------------+--------------
ward a | cllr 1 | cllr 2 | cllr 23 |
ward b | cllr 4 | cllr 5 | cllr 8 |
(2 rows)
It is all described here: http://www.postgresql.org/docs/9.1/static/tablefunc.html#AEN142967
Upvotes: 1
Reputation: 1269503
I tend to approach this type of query using conditional aggregation:
select ward,
max(case when seqnum = 1 then councillor end) as councillor1,
max(case when seqnum = 2 then councillor end) as councillor2,
max(case when seqnum = 3 then councillor end) as councillor3
from (select wc.*,
row_number() over (partition by ward order by councillor) as seqnum
from ward_councillors wc
) wc
group by ward;
Upvotes: 4