Wikistats Overview

This page is a concise overview of Wikimedia stats reports built (almost all) by Erik Zachte, and often referred to as Wikistats.
This software has its own git repository, see GitHub analytics-wikistats.

BTW the term Wikistats is used differently in different contexts. The website stats.wikimedia.org features much more software, developed by other Wikimedia staff and volunteers.

Four sets of reports come in (almost) independent clusters.

Cluster	I Dump reports	II Page view reports	III Squid reports	IV Mail stats
Focus	Wiki content and activity	Traffic	Traffic	Mailing list activity
Input	Wikimedia dumps Only dumps, which contain all historic revisions. Mostly stub dumps = meta data only Sometimes full archive dumps (meta data + raw text)	Hourly page view counts per wiki (BTW also aggregated into daily/monthly dumps)	Squid logs, 1:1000 sampled	Master list Mail archive per list
Output	Per project (8, plus misc.) a sitemap, e.g. Wikipedia (Wp) For each wiki a report with 10+ tables, e.g. English Wp Per project 18 comparison reports, e.g. Edits per month on Wp Current status all wikis, e.g. Wp Bot created/edited articles per wiki per month, e.g. Wp Overview recent months, e.g. Wp Largest/most edited articles, e.g. (all in all ± 800 wiki specific pages, plus 144 comparisons, each in 28 languages -> ± 25,000 static html pages )	Page views per month: per site (mobile, main, combined), per language, raw or normalized, grouped by project/region As above, overview over all projects (unofficial) Page views by article for a selection of languages, and a selection of category trees	Page views/edits per country and/or per wiki, plus trends (Quarterly/Yearly) Breakdown of server requests by referrer, target wiki, mime type, browser/app (mobile/non-mobile), operating system (mobile/non-mobile), Breakdown of server requests from or via Google search/apps	Activity per mailing list Power posters
Output Languages	28 ¹	English	English	English
Updated	Monthly	Daily	Monthly	Daily
Runs on	WMF server	WMF server	WMF server	private server
Code base	Dumps	Dumps it reuses reporting code built for dumps	Squids	Mail-lists
Non blank perl lines	57,539	4,898 ²	14,483	1,227
Dev languages	Reports: perl Charts: R Animation: html5+javascript	Reports: perl	Reports: perl Animation: html5+javascript Geo2ip: MaxMind	Reports: perl
Feed to Wikimedia Report Card, based on Limn via Analytics	Yes	Yes	No	No
Animations	Bubble charts code	n.a.	Interactive map code	n.a.
Monitoring	Dumps generation official Dumps generation concise code (slow response) Dumps processing code	On udp2log level, udp packet loss stats: Realtime, aggregated Realtime, per server, e.g. gadolinium Monthly, per server location / server role		No
Examples of charts
Author	Erik Zachte 2003-now	Erik Zachte 2008-now	Erik Zachte 2009-now Andre Engels 2012	Erik Zachte 2005?-now
Notes	¹ Some reports only in English, most translations are incomplete. ² For data collecting and archiving. Reporting code is part of dumps scripts.

Last upd: June 25, 2013

Wikistats Overview

I Dump reports

II Page view reports

III Squid reports

IV Mail stats