I'm using 16/02/12 21:28:38 INFO Worker: Connecting to master pl:7077...
16/02/12 21:28:38 INFO Worker: Successfully registered with master spark://pl:7077
on postgres. I have a query which goes like this:
RedshiftHere, I'm selecting a combination of SELECT EXTRACT(year from created_at) AS CustomYear,
client_ip,
member_id,
COUNT(*) AS Views
FROM ads.fbs_page_view_staging
WHERE member_id = 2
GROUP BY CustomYear,
client_ip,
member_id
HAVING COUNT(*) = 1
ORDER BY CustomYear
and client_ip where member_id is 1. I would now like to take these combinations of Views and client_ip and subset the entire table member_id having only such combinations.
If there was only one column I wanted to subset on, say ads.fbs_page_view_staging, I could've written the following query and got the results:
client_ipNotice that in the outer query, I am subsetting based on SELECT EXTRACT(year FROM created_at) AS CustomYear,
COUNT(*)
FROM ads.fbs_page_view_staging
WHERE member_id = 2
AND client_ip IN (SELECT client_ip
FROM ((SELECT EXTRACT(year from created_at) AS CustomYear,
client_ip,
member_id,
COUNT(*)
FROM ads.fbs_page_view_staging
WHERE member_id = 2
GROUP BY CustomYear,
client_ip,
member_id
HAVING COUNT(*) = 1
ORDER BY CustomYear)))
GROUP BY customyear
ORDER BY customyear
. But how do I subset the table on a combination of columns?
Any help would be much appreciated.
答案 0 :(得分：0)
Instead of subquerying, try joining directly to the results of your query. That way you can specify multiple criteria.
Here is (draft) SQL to select IP/member pairs that match the rows found by your sub-query (i.e. for some year in the past, there was only one view for that IP & member.)
faces = face_cascade.detectMultiScale(gray, 1.3, 5)
I'm not certain I've captured the intent of your query correctly, but if not hopefully you can adapt the technique.