ngkountas opened https://gitlab.wikimedia.org/repos/lpl/starterkit-v1/-/merge_requests/15

Dashboard: Add first version of "Activity & Growth" card

toluayo merged https://gitlab.wikimedia.org/repos/lpl/starterkit-v1/-/merge_requests/15

Dashboard: Add first version of "Activity & Growth" card

ngkountas opened https://gitlab.wikimedia.org/repos/lpl/starterkit-v1/-/merge_requests/16

Dashboard "Activity & Growth": Add actual values for new vs deleted articles

toluayo merged https://gitlab.wikimedia.org/repos/lpl/starterkit-v1/-/merge_requests/16

Dashboard "Activity & Growth": Add actual values for new vs deleted articles

Summarizing the discussion with @KCVelaga_WMF and @ngkountas offline about the metrics to use for this component.

We discussed that besides active editors, rest of the metrics (new articles count/deleted articles count, top reader topics, top editor topics) are a bit tricky to fetch. Upon investigation, @ngkountas found that that we can fetch the number of new pages easily as there is already an API endpoint for that. (example for tdd wiki) and that we can also rely on the logevents API to fetch the deleted pages and count them manually (example for tdd wiki).

About the top read / edited pages and their respective topics as per @KCVelaga_WMF, the output can be Top pages by pageviews:

Page A - biography
Page B - sports
Page C - history

So we'd get the most viewed pages (API endpoint), and get topics from the article topics model. But that won't be what topics are read most, but rather what topics the top read pages belong to.

If we want to fetch most read topics, then we'd have get topics for all pages on the wiki, sum by topic and get the top N topics. Topic areas is a bit of rabbit hole. The model predicts multiple topics with probability. The end user has to decide the threshold based on their application.

We have pageview data, and article topics model output. But currently, we don't have a dataset that combines these both. It is technically possible, but since there hasn't been a regular need for it, there is no pipeline or API readily available. Creating a new dataset might be out of scope here.

@ngkountas To me it sounds like based on the alternative @KCVelaga_WMF has provided for 3rd and 4th metrics we can show top read / edited pages and their respective topics. I think if the API for topics is not available it is fine to show what topics the top read/edited pages belong to?

API for article topics model output: https://api.wikimedia.org/wiki/Lift_Wing_API/Reference/Get_articletopic_outlink_prediction

ngkountas opened https://gitlab.wikimedia.org/repos/lpl/starterkit-v1/-/merge_requests/32

Complete ActivityGrowthCard basic functionalities

eamedina merged https://gitlab.wikimedia.org/repos/lpl/starterkit-v1/-/merge_requests/32

Complete ActivityGrowthCard basic functionalities

ngkountas opened https://gitlab.wikimedia.org/repos/lpl/starterkit-v1/-/merge_requests/44

Add chevron icon in Activity & Growth stat cards

toluayo merged https://gitlab.wikimedia.org/repos/lpl/starterkit-v1/-/merge_requests/44

Add chevron icon in Activity & Growth stat cards

While testing the activity & growth feature with test.wikipedia, I noticed a few things:

ACTIVE EDITORS

• I am not seeing the data for the month of April. I am assuming the data is not yet available, which is why it is missing?
• I wonder how comments are generated such as "Declining — focus on community building" for editor count.
• I am not sure what "~32 this month" means. Looking at the mockups may be it is a "+"

NEW VS. DELETED

• In some cases where the number is greater than two digits, it is overlapping with other numbers.
• Same comment as above for this section as well - I do not see data for April here.
• It looks like we are showing monthly data in all sections. Probably we could tweak "Track your wiki's health and progress toward becoming a Small Wiki" slightly to "Track your wiki's monthly health and progress toward becoming a Small Wiki".

TOP EDITOR TOPICS

• I see this empty right now for test.wikipedia.org.

ROADMAP TO NEW WIKIS
I am guessing this is not taking into account the number of articles the wiki has. It is probably not a big deal for now, but I am wondering that if most wikis invited for the pilot are already small wikis, this roadmap may not be very useful for them. So, going by the definition of small and large wikis on this page https://meta.wikimedia.org/wiki/Small_and_large_wikis#Large_wikis, perhaps we could introduce a milestone for small wikis as well, for example 50,000 articles to become a large wiki.

Overall, I think we could remove em dashes "—" in the UI. For example:

• 🎯 Next milestone: articles, 0 to go
• 📉 Declining, focus on community building

Thank you for your comments @srishakatux! I'll address them below.

ACTIVE EDITORS

I am not seeing the data for the month of April. I am assuming the data is not yet available, which is why it is missing?

Your assumption is right, sometimes the data for the previous month is not available.

I wonder how comments are generated such as "Declining — focus on community building" for editor count.

The difference between the two last months is negative, that how we deduct that the editors count is declining.

I am not sure what "~32 this month" means. Looking at the mockups may be it is a "+"

The sign used here is minus (-). The editor count in the last (available) month was 58, while the previous month it was 90. So the difference is (-32).

NEW VS. DELETED

In some cases where the number is greater than two digits, it is overlapping with other numbers.

Good catch, needs to be fixed, but is not of high priority, as the small wikis we are targeting are not expected to have number with more than two digits.

Same comment as above for this section as well - I do not see data for April here.

Same response as above.

It looks like we are showing monthly data in all sections. Probably we could tweak "Track your wiki's health and progress toward becoming a Small Wiki" slightly to "Track your wiki's monthly health and progress toward becoming a Small Wiki".

We can fix this.

TOP EDITOR TOPICS

I see this empty right now for test.wikipedia.org.

Most edited pages in test wiki are mostly pages in the test namespace. Even for pages that are in the main namespace, the articletopics API doesn't return any topic. In order to test this card properly, you can use an actual production wiki.

ROADMAP TO NEW WIKIS

I am guessing this is not taking into account the number of articles the wiki has. It is probably not a big deal for now, but I am wondering that if most wikis invited for the pilot are already small wikis, this roadmap may not be very useful for them. So, going by the definition of small and large wikis on this page https://meta.wikimedia.org/wiki/Small_and_large_wikis#Large_wikis, perhaps we could introduce a milestone for small wikis as well, for example 50,000 articles to become a large wiki.

That's a good idea, which would also fix the same issue in header. Please consider creating a new phabricator task to track this issue.

ngkountas opened https://gitlab.wikimedia.org/repos/lpl/starterkit-v1/-/merge_requests/81

Fix: bound deleted articles log query to the requested date range

ngkountas opened https://gitlab.wikimedia.org/repos/lpl/starterkit-v1/-/merge_requests/82

Fix: bound deleted articles log query to the requested date range

eamedina merged https://gitlab.wikimedia.org/repos/lpl/starterkit-v1/-/merge_requests/81

Fix: bound deleted articles log query to the requested date range

toluayo merged https://gitlab.wikimedia.org/repos/lpl/starterkit-v1/-/merge_requests/82

Fix: only return months present in both article trends