heuristic to find wrong coordinates and / or wrong iso-codes

This query is marked as a draft This query has been published by Herzi Pinki.

Find coordinates by iso-code that are far away from the average center of all coordinates with the same iso-code and sort them by descending distance from that center. This is a heuristic to find wrong coordinates and / or wrong iso-codes. If you get to the first point already inside the iso-code area, it is unlikely to find more errors further down below. Works best for iso areas that are close to cycle shape (Switzerland, as opposed to the USA), and it does not work for areas located on both sides of the 180° meridian (e.g. FJ, NZ).
Fix & rerun. You can change iso codes through variables iso1 and iso2 (you have to copy the script).
Original script by Thomas Ledl.

SQL

use dewiki_p;
set @iso1:='BW';
set @iso2:='19';

select sqrt((gt_lat - cen_gt_lat)*(gt_lat - cen_gt_lat) + (gt_lon - cen_gt_lon)*(gt_lon - cen_gt_lon)) as distanzungefaehr, 
       gt_lat, gt_lon, gt_page_id, gt_name,
       (select page_title from page where page_id = gt_page_id) as title,
       gt_country,gt_region from geo_tags,
       (select avg(gt_lat) cen_gt_lat,avg(gt_lon) cen_gt_lon,count(*) from geo_tags
        where gt_country=@iso1 /*and gt_region = @iso2*/) as zentrumdavon
       where gt_country=@iso1 /*and gt_region = @iso2*/
       order by 1 desc limit 100;

​x
 
use dewiki_p;set @iso1:='BW';set @iso2:='19';​select sqrt((gt_lat - cen_gt_lat)*(gt_lat - cen_gt_lat) + (gt_lon - cen_gt_lon)*(gt_lon - cen_gt_lon)) as distanzungefaehr,        gt_lat, gt_lon, gt_page_id, gt_name,       (select page_title from page where page_id = gt_page_id) as title,       gt_country,gt_region from geo_tags,       (select avg(gt_lat) cen_gt_lat,avg(gt_lon) cen_gt_lon,count(*) from geo_tags        where gt_country=@iso1 /*and gt_region = @iso2*/) as zentrumdavon       where gt_country=@iso1 /*and gt_region = @iso2*/       order by 1 desc limit 100;

By running queries you agree to the Cloud Services Terms of Use and you irrevocably agree to release your SQL under CC0 License.

Submit Query Stop Query

All SQL code is licensed under CC0 License.

Query status: complete

Executed on Thu, 27 Oct 2016 07:39:02 UTC.