Comments on: SQL CHALLENGE http://blog.johnjosephbachir.org/2007/01/19/sql-challenge/ John Joseph Bachir's blog Sat, 30 Aug 2008 01:19:30 +0000 http://wordpress.org/?v=2.6.1 By: John http://blog.johnjosephbachir.org/2007/01/19/sql-challenge/#comment-112510 John Tue, 23 Jan 2007 22:49:38 +0000 http://blog.johnjosephbachir.org/2007/01/19/sql-challenge/#comment-112510 Here is something else I came up with, which works properly on some datasets, but when I tried it where every row had the same zip, it came up with 0 rows, not 1. But anyway, it's my current best solution. <code>SELECT DISTINCT * FROM users u1, users u2 where u1.zipcode != u2.zipcode ORDER BY RAND(UNIX_TIMESTAMP());</code> Here is something else I came up with, which works properly on some datasets, but when I tried it where every row had the same zip, it came up with 0 rows, not 1. But anyway, it’s my current best solution.

SELECT DISTINCT * FROM users u1, users u2 where u1.zipcode != u2.zipcode ORDER BY RAND(UNIX_TIMESTAMP());

]]> By: John http://blog.johnjosephbachir.org/2007/01/19/sql-challenge/#comment-111327 John Fri, 19 Jan 2007 20:02:57 +0000 http://blog.johnjosephbachir.org/2007/01/19/sql-challenge/#comment-111327 Whoah, I just tried it and verified that you are right. Well that makes for a more simple query to start out with. Hmmm Whoah, I just tried it and verified that you are right. Well that makes for a more simple query to start out with. Hmmm

]]> By: Jason http://blog.johnjosephbachir.org/2007/01/19/sql-challenge/#comment-111326 Jason Fri, 19 Jan 2007 19:56:43 +0000 http://blog.johnjosephbachir.org/2007/01/19/sql-challenge/#comment-111326 Actually no. using having count(zipcode) >=1 means it'll return anything that has at least 1 zipcode in it. It doesn't do much for filtering in this case. Actually no. using having count(zipcode) >=1 means it’ll return anything that has at least 1 zipcode in it. It doesn’t do much for filtering in this case.

]]> By: John http://blog.johnjosephbachir.org/2007/01/19/sql-challenge/#comment-111207 John Fri, 19 Jan 2007 17:18:45 +0000 http://blog.johnjosephbachir.org/2007/01/19/sql-challenge/#comment-111207 Hey, thanks for the attempt danny, but that still returns the same row every time :( and now i am even more confused, because I thought that <code>HAVING count(zipcode)>=1</code> was the part that squeezed the GROUP BY groups (each zip code cluster) down to 1. Hey, thanks for the attempt danny, but that still returns the same row every time :(

and now i am even more confused, because I thought that HAVING count(zipcode)>=1 was the part that squeezed the GROUP BY groups (each zip code cluster) down to 1.

]]> By: danny http://blog.johnjosephbachir.org/2007/01/19/sql-challenge/#comment-111192 danny Fri, 19 Jan 2007 17:06:34 +0000 http://blog.johnjosephbachir.org/2007/01/19/sql-challenge/#comment-111192 Try this (tweaked since UNIX_TIMESTAMP() is optional, and WHERE is quicker than HAVING): SELECT * FROM USERS WHERE zipcode > '' GROUP BY zipcode ORDER BY zipcode, RAND(); Try this (tweaked since UNIX_TIMESTAMP() is optional, and WHERE is quicker than HAVING):

SELECT * FROM USERS WHERE zipcode > ” GROUP BY zipcode ORDER BY zipcode, RAND();

]]>