Page MenuHomePhabricator

Provide stats for visitors to Wikimedia project (www) portals
Closed, ResolvedPublic

Description

https://meta.wikimedia.org/wiki/Project_portals

We have project portals such as https://www.wikipedia.org and https://www.wiktionary.org. It'd be nice to get the raw data of number of visitors on an hourly basis to the various project (www) portals.

I believe the Wikimedia Analytics team is continuing to look for discrete projects to work on. Toby: thoughts on this being a 2014Q1 analytics project?


Version: unspecified
Severity: enhancement
URL: https://meta.wikimedia.org/wiki/Project_portals

Details

Reference
bz57828

Event Timeline

bzimport raised the priority of this task from to Low.Nov 22 2014, 2:36 AM
bzimport set Reference to bz57828.
bzimport added a subscriber: Unknown Object (MLST).

"numbers of visitors" are rather hard with our current infrastructure.
We discussed in IRC that "numbers of requests" might be ok as well.

"number of requests" would be easier, as we could just filter the
sampled-1000 stream ... which would be fairly straight-forward.

Extracting hourly breakdowns should not be much hassle.

We could for example just put them in a TSV and put them in a public
place.

Monitoring would be easy as well ...

If such a solution would help you ... one could realize that fairly easily.

But as we're severly understaffed, I'll have to defer to Toby whether that
would be interesting for Analytics, and when/if he can sneak that in
somewhere in the Analytics schedule.

As Christian pointed out, this is not particularly difficult. But I'd prefer to wait until our new processing infrastructure is in place. I'll post some more information on this to the Analytics list this week.

-Toby

It's here: https://stats.wikimedia.org/wikimedia/squids/SquidReportPageViewsPerLanguageBreakdown.htm#Portal

Portal (www) (3.3% share of global total)
United States 38.8%
India 21.1%
Canada 3.8%
United Kingdom 3.0%
Unknown 2.2%
[...]