Create a map showing where we have a higher proportion of uploads than other upload tools

[nicolas-raoul]

I believe our app results in more uploads from the global South / underrepresented countries than the web Upload Wizard, but I don't think we have a solid proof of that.

This script could be customized to check for our label, then compare. For instance it could show with a green gradient the areas where we have proportionally more than our average share of uploads :

# Quad-licensed under CC-BY-SA 3.0, GNU AGPL v3, MIT/Expat License, NCSA Open Source License
# Use and distribution are permitted as long as any one of the above licenses is satisfied.
from PIL import Image, ImageDraw

im = Image.new("L", (720, 360))
draw = ImageDraw.Draw(im)

query = """
SELECT 180-ROUND(gt_lat*2),
    ROUND(gt_lon*2)+360,
    ROUND(LOG2(COUNT(gt_page_id))*16)
FROM geo_tags
INNER JOIN page
    ON page_id = gt_page_id
WHERE gt_globe = "earth"
    AND NOT (
        gt_lat = 0
        AND gt_lon = 0
    )
    AND page_namespace = 6
GROUP BY ROUND(gt_lat*2),
    ROUND(gt_lon*2)
ORDER BY COUNT(gt_page_id) DESC
;"""

import MySQLdb

connection = MySQLdb.connect(
    host = "commonswiki.labsdb",
    db = "commonswiki_p",
    read_default_file = "~/replica.my.cnf"
)
cursor = connection.cursor()
cursor.execute(query)

for y, x, c in cursor.fetchall():
    draw.point((x,y), fill=c)

im.save("File:Geolocated_images_in_Wikimedia_Commons_2014-02-21.png", "PNG")

Found at https://commons.wikimedia.org/wiki/File:Geolocated_images_in_Wikimedia_Commons_2014-02-21.png#Generator

Anyone knows whether there is a lightweight way to execute this script, in particular without setting a lab instance and downloading the database? Maybe Quarry? As usual, all volunteers welcome :-)

[misaochan]

Great idea! I can obtain maps of active installs from the dev console, but uploads is a better metric for sure.