Talk:Pageviews Analysis/Archives/2019/1

From Meta, a Wikimedia project coordination wiki

2017 yearly report for Armenian Wikipedia is the same as 2018 yearly report

Resolved.

Hi, please check this and this links. I think there is an error here, it is almost impossible to have exactly the same result in different years.--ԱշոտՏՆՂ (talk) 16:33, 5 January 2019 (UTC)

@ԱշոտՏՆՂ: Fixed. Sorry about that! There was an error in the script I used to import the data. Best, MusikAnimal (WMF) (talk) 16:45, 5 January 2019 (UTC)

External links Massviews shouldn't default to HTTP only

By default - if you search for *.website.com - Special:LinkSearch only searches HTTP links (see Phab:T14810). While Massviews defaulting to LinkSearch behaviour when searching for external link pageviews is understandable, I'd really like to be able to get statistics for both HTTP and HTTPS through the tool without making separate queries. Would it be possible to, at least optionally, search for both HTTP and HTTPS links when entering a generic search pattern? Samwalton9 (WMF) (talk) 12:12, 11 January 2019 (UTC)

Cache Massviews queries for large Categories - Feature Request

I think that Massviews queries for large categories like Category:Wikipedia introduction cleanup > 1000 pages should be cached daily for faster loading and to reduce server stress, because large categories like these get queried often in the Community Portal in order to find popular pages to edit. Categories with >10,000 pages (including subcategories) take a minute or more to load. Turbo Ink (talk) 16:55, 5 January 2019 (UTC)

We could certainly add some time-based caching, but it would only happen after the initial request is made. For example, you search for Category:Wikipedia_introduction_cleanup, and then everyone else who searches it in say, the next hour, will get the cached results. I think 24 hours may be too long. MusikAnimal (WMF) (talk) 20:33, 13 January 2019 (UTC)
Yes, that's what I had in mind. Caching large requests would be good whether the cache lasts for an hour or a few hours. Turbo Ink (talk) 07:35, 15 January 2019 (UTC)

number of reverts on page

Would you please add a column to show the number of reverts and rollbacks on the pages? for example at here after the number of editors Yamaha5 (talk) —The preceding comment was added without date/time by Yamaha5 (talk) 19:00, 5 January 2019 (UTC)

This is a fun idea, but I think it might slow down the tool quite a bit. Things like partial reverts can also be undetectable. I will probably have to decline this on the basis of engineering cost/benefit and performance implications. Sorry! MusikAnimal (WMF) (talk) 20:42, 13 January 2019 (UTC)

Q7403848 article in Arabic Wikipedia

Resolved.

Hello, I opened this section based on this discussion. سليم دبور article received (and still) unexplained views. Since 10 Dec 2018 we started search about explanation but we can't found anything! So, is it normal?! --Alaa :)..! 15:07, 12 January 2019 (UTC)

@علاء: Hey! Sorry for the late reply. I looked into this, and there are some anomalies: for starters, it appears that viewership on your wiki is usually at least 75% mobile traffic. ar:سليم دبور however received around 35% mobile viewership in December 2018 [1]. So that to me is a red flag. I then queried the Analytics database to see where traffic originated from. I looked at days where there was a spike, specifically December 1 and December 13. On both days, a very high percentage came from a single city, which is the tale-tale sign of false traffic. However, December 1 and December 13 showed different cities. This is a bit unusual; if there's ongoing false traffic, it usually is always from the same city. I won't mention which cities these are, but they are not big cities. In fact, if each pageview counted as a different person (it doesn't, but let's pretend it does), it'd be a sizable portion of the entire population of those cities. That said, it is my opinion that most traffic to سليم دبور is probably an undeclared bot and not legitimate, human traffic. Why someone would want a bot to scrape this page is anyone's guess :) Undeclared bots are common, just so you know. More info at [2]. Hope this helps, MusikAnimal (WMF) (talk) 21:21, 14 January 2019 (UTC)
@MusikAnimal (WMF): Thanks a lot, I'll pass your comment to arwiki community. Best --Alaa :)..! 18:04, 18 January 2019 (UTC)
@MusikAnimal (WMF): can we avoid this page in arwiki views? as it still appear until now! --Alaa :)..! 02:23, 23 May 2019 (UTC)
@علاء: I've added سليم دبور to the blacklist, so it won't show up in Topviews anymore. PS -- we shouldn't be talking in the talk page archives! :) MusikAnimal (WMF) (talk) 05:18, 28 May 2019 (UTC)
Thanks @MusikAnimal (WMF):. Yes ofc I know that we shouldn't talking in take pages, but as there's some stalkers I ping you here. Best --Alaa :)..! 05:21, 28 May 2019 (UTC)

Roach

Hello. I'm trying to see the number of views for ru:Роуч but it doesn't appear in the results when I type its name in, unlike the other pages with names starting with Роуч. Why could that be? My browser is Opera. WikiArticleEditor (talk) 21:35, 13 January 2019 (UTC)

@WikiArticleEditor: Seems like an issue with the autocompleteion. I will look into it! For now, you can get to it by either (a) putting the article directly in the URL, or (b) go to Settings and choose "No autocompletion" for the "Search method". Here are the results for the last 20 days: [3] MusikAnimal (WMF) (talk) 21:47, 13 January 2019 (UTC)
Thanks. WikiArticleEditor (talk) 22:06, 13 January 2019 (UTC)

Redirects

Hello I can not find old stats for redirected web page. Now (since 21 december 2018):

  • Zespół pałacowo parkowy w Wodzisławiu Śląskim-Kokoszycach (all ok)

Untill 20 december 2018:

  • Zespół pałacowo parkowy w Wodzisławiu Śląskim - Kokoszycach (stats gone)

--Swd (talk) 07:10, 20 January 2019 (UTC)

"All languages" links do not work when starting from Wikidata

I was looking at this set of Wikidata items and wanted to see "All languages" data for them, yet the links did not work, e.g. for Q33, I am getting an error "Q33 - No data found." -- Daniel Mietchen (talk) 23:58, 22 January 2019 (UTC)

Massviews Hashtags needs to be pointed to the new tool

The Hashtags option in Massviews is still using the old Toolforge-hosted tool. The tool has since been re-written and is now available at https://hashtags.wmflabs.org/. JSON responses can be retrieved by the URL structure https://hashtags.wmflabs.org/json/?query=<hashtag>&project=<project url>&startdate=<YYYY-MM-DD>&enddate=<YYYY-MM-DD> e.g.https://hashtags.wmflabs.org/json/?query=test&project=en.wikipedia.org&startdate=2018-01-01&enddate=2019-01-05 :) Samwalton9 (WMF) (talk) 15:57, 8 February 2019 (UTC)

en.wikiversity

Pageviews Analysis no longer works for en.wikiversity pages. Is there anything we can do to correct this? -- Dave Braunschweig (talk) 18:40, 12 February 2019 (UTC)

@Dave Braunschweig: Could you give me an example? It appears to be working for me, e.g. [4]. Please also let me know what web browser you are using. MusikAnimal (WMF) (talk) 18:56, 12 February 2019 (UTC)
@MusikAnimal (WMF): Very interesting. Your link didn't work in Chrome, but I can access en.wikipedia stats correctly from the same browser and screen. If I switch to Safari, it is working correctly for en.wikiversity. Clearly, it is something installed or configured on my end. I'll keep experimenting. Thanks for following up! -- Dave Braunschweig (talk) 20:47, 12 February 2019 (UTC)
@Dave Braunschweig: Do you by chance use a anti-tracking browser extension, such as Privacy Badger? Many of these will by default block any requests to a Wikimedia API. So you would need to whitelist en.wikiversity in addition to en.wikipedia, etc. MusikAnimal (WMF) (talk) 21:58, 12 February 2019 (UTC)
@MusikAnimal (WMF): That was it. I adjusted Privacy Badger and all is well. Thanks for your help! -- Dave Braunschweig (talk) 23:27, 12 February 2019 (UTC)

Timezone / Timeframe

This question is about the PageViews API, but it applies to all APIs.

Basically I was wondering which timezone is used to decide what a day is, and if it differs from project to project (e.g. en.wikipedia vs de.wikipedia). There was no real information on that in the FAQ or documentation.

I am assuming you are either using UTC or the timezone configured in your WMF Hadoop data lake (which might be UTC). Can you please clarify on that? :-)

FranzJUniv (talk) 16:44, 18 February 2019 (UTC)

@FranzJUniv: I am quite certain everything is in UTC. This applies to all APIs. MusikAnimal (WMF) (talk) 21:50, 18 February 2019 (UTC)

Addition of wikimaniawiki?

Tracked in Phabricator:
Task T216525

Are all the wikis going to end up with analysis by default? Or is there the need to request their addition?  — billinghurst sDrewth 05:09, 19 February 2019 (UTC)

@Billinghurst: Yes unfortunately it does need to be requested on a per-wiki basis. I have done so at phab:T216525. Updating Pageviews Analysis is actually a separate process, but I usually am diligent about staying in sync with the underlying API. Best, MusikAnimal (WMF) (talk) 16:53, 19 February 2019 (UTC)
@MusikAnimal (WMF): Thanks for your answer. I am guessing that this is one wiki where there will be lots of mobile views, especially at Wikimanias, and it will be a decent measure to watch traffic with regard to announcements.  — billinghurst sDrewth 22:09, 19 February 2019 (UTC)

No hay estadísticas de los artículos exclusivamente en español

No existen estadísticas sobre las páginas en español que no tienen versiones en otros idiomas (o por lo menos en ingles). Por ejemplo, no logro encontrar las visitas del artículo "Rafael Núñez (goberador). Saludos. — The preceding unsigned comment was added by 190.106.110.83 (talk) 20:48, 20 February 2019 (UTC)

Hola, podrías darme un enlace, por favor? Esta trabajando para mi, por ejemplo [5]. ¿Tal vez estabas hablando de Langviews? De hecho, no hay estadísticas para Rafael Núñez (goberador), porque no hay artículos en otros idiomas (mira wikidata:Q6097752). Esto es esperado. Si no te entiendo, por favor házmelo saber. Espero que esto te ayude. Saludos, MusikAnimal (WMF) (talk) 21:15, 20 February 2019 (UTC)

Add option "Copy names to clipboard"

Hello,

Could you add to lists an option "Copy names to clipboard", so we can use this list in petscan. Now I download list in csv and then copy names to petscan. --Alex Blokha (talk) 00:15, 27 February 2019 (UTC)

Showing data values in all the graph

Hi! Good job with this tool, congratulations! I've seen that in some variants of the graphs, it's not possible to show the total data values. Could you include the option to integrate them, as in the line chart? Thank you! Xavi Dengra (MESSAGES) 10:10, 4 March 2019 (UTC)

Default should be "Begin at zero" IMO

People can click it of if they like. That will give a more accurate initial visual of pageviews over time. Best Doc James (talk · contribs · email) 07:31, 5 March 2019 (UTC)

@Doc James: You can set this preference by going to Settings and checking "Always show the y-axis starting at zero" under "Chart preferences". The idea behind the default behaviour is that you can more easily see fluctuations in the data. Many pages have a regular pattern that may otherwise be obfuscated. MusikAnimal (WMF) (talk) 20:12, 9 March 2019 (UTC)
User:MusikAnimal (WMF) thanks. Did not see that I could set it to stay with starting at zero. Doc James (talk · contribs · email) 02:37, 10 March 2019 (UTC)

Yellow color for 9th page; need dark mode

Very often I could not see the yellow color for a 9th page very well in Line, Bar, and Radar chart types. Is the yellow color harder to see? The background is white; the yellow color looks too obscure. Here's my sample.

I would like to request "Dark mode" for the WMFLabs website. That way, colors would be much easier to see in the dark background, especially the yellow color. George Ho (talk) 03:26, 12 March 2019 (UTC)

An extraordinarily attractive article ?

Two million pageviews in February 2019 of the rather minor article de:Anthocyane on water-soluble pigments and close to one million pageviews for the last 30 days? Second only to the German Wikipedia main page? Certainly a glitch in the software, isn't it?

best regards, — The preceding unsigned comment was added by KaiKemmann (talk)

@KaiKemmann: This is likely an undeclared bot. A quick way to check such anomalies is to compare desktop and mobile pageviews. You'll see that in this case, the spike in pageviews was only on desktop. Mobile web and mobile app look normal. The Anthocyane page was removed from the Topviews results for this reason. Why a bot would scrape this page is anyone's guess. Hope this helps, MusikAnimal (WMF) (talk) 21:53, 26 March 2019 (UTC)

Potential Bug Report

A page that has existed for several years has had its pageview statistics, other than pageviews within the last few months, somehow deleted. This problem is experienced while using Chrome on Windows 10.

https://en.wikipedia.org/wiki/Mike_Dunleavy_(politician) — The preceding unsigned comment was added by 158.145.240.112 (talk)

The page was moved in November 2018. Unfortunately the pageviews don't move with it. A redirect was left, so you can get combined pageviews by using the Redirect Views tool. When using Pageviews Analysis, you can get to Redirect Views using the "Redirects" link below the stats in the right column. MusikAnimal (WMF) (talk) 21:48, 26 March 2019 (UTC)

Mediaviews bug report

  1. link to break
  2. on enter, hangs without returning any results. Time circle is seen
  3. Chrome, Win7 Pro. Ian Furst (talk) 15:56, 28 March 2019 (UTC)
Thanks for the detailed report! Unfortunately the external service this tool relies on is currently down. The maintainers are aware and are working to restore service. Sorry for the disruption, MusikAnimal (WMF) (talk) 16:12, 28 March 2019 (UTC)