Talk:Wikistats 2.0 Design Project/RequestforFeedback/Round1/Detail page single wiki

Wikistats 2.0 Design Project/RequestforFeedback/Round1#Single Wiki describes our design thinking for this part of Wikistats 2.0. This board is meant to collect feedback from current and potential future users of this community tool.

What formats would you like besides CSV, TSV, and JSON?

Is it possible to have good old tables for all wikis available somewhere? This is the simplest way for using data. It would be great to be able to download tables like at https://stats.wikimedia.org/EN/TablesPageViewsMonthlyCombined.htm , without having to download wiki per wiki

Otherwise yes, it is really important (at least for me) to be able to download data. For instance, recently I made a study to find out what impacts page views more: increase in bot edits or increase in number of human active users? This was quite simple as I could download (a) pageview table for all wikis, (b) active users table for all wikis, (c) bot edits % for all wikis, (d) article counts for all wikis. Then it was quite straightforward as I had to process just 4 spreadsheets and could select most prominent examples (e.g. wikis with most bot edits, wikis with highest page view growth) for each. It would have been way more difficult if I had had to download wiki per wiki.

To sum up yes, easy downloading is essential for me.

Good to know. So all-wiki data is important, and access to the raw data is important. We were talking about having an "advanced" interface for these kinds of use-cases, so we can keep the first thing people see more approachable. Do you have use-cases that would benefit from the simpler interface, or most of your work is cross-wiki?

Yes, of course. I cited them in some other section (sorry, can't find it with Flow, some Topic:Tkaqwertyuiop...): I also look at simple indicators, like increase in number of active users on Commons during Wiki Loves Earth, or increase in new articles on one particular Wikipedia during an edit-a-thon. Graphs are useful for such indicators.

Oh, and one more example: I wanted to find the number of active users for all projects per language and find most active languages per region. Once again, straightforward if you can download data (sum of few tables plus regions from https://stats.wikimedia.org/EN/Sitemap.htm ), but hardly possible without downloading option.

We are not Google who hides trends data and shows only relative graphs for Google Trends. Please make data available for download like one could easily copy old tables. Or keep old tables (no matter they are ugly, having data is essential).

It is important, specially for research purposes and comparative studies.

Let me testify that this feature was asked a lot for Wikistats 1.0 and I mean really a lot. Many people wanted to play with data themselves. Sadly it never happened, as it would either complicate the code further or require a major rewrite. Wikistats 1.0 does data aggregation and presentation in one go (not a lucky choice), so code is already interspersed with output statements. Writing a flat csv file besides a not quite flat html table would make the code even harder to maintain.

Currently considered are all, year, 3-month, and month.

All time is really needed. Would be great to have an adjustable range (e.g. I want to see development of Ukrainian Wikisource from 2006 to 2008 only)

I would add to the ones provided: 6 months.

Wikistats page views uses 24 months. I'd say 24,36,48 or even 60 months instead of 12 months helps greatly to not draw wrong conclusions from growth or decline where these are essentially (partially) random fluctuations. Easy to check at https://stats.wikimedia.org/EN/TablesPageViewsMonthlyCombined.htm Imagine only last 12 months were shown. In many cases trend would seem consistently up or down where wider scope shows this is not the case at all.

Do you find one of line, bar, area, or other charts particularly enlightening? Do you often need to see data in different scales (logarithmic, indexed, or other)?

Bar chart seems fine. I would add line if possible, especially if you need to compare two different metrics.

We display a metric and all available breakdowns in a View. Is this terminology intuitive and is the navigation structure under a Topic clear?

It is clear, but I wouldn't call it "Views". I would call it "Metric".

If so, please provide a concrete example.

It is quite useful to compare # of active users at different levels of activity (1+, 5+, 100+ etc.) or page views of different types (mobile / desktop / total) in the same visualisation. It is OK not to implement it if easy downloading is implemented.

+ 1 to NickK

Also, it is useful to compare newly registered editors and editors retained.

...

I can look at graphs to find interesting metrics. I can also use them to illustrate reports (e.g. impact of Wiki Loves Earth on # of active users on Commons). I would use it way more often if it would contain values (tables) instead of graphs

Really good to know. So given that you prefer tables, would you like the site to remember this preference and just render everything as tables if possible by default?

Actually both can work, I have no particular preference. If most people prefer graphs but I will have to make an extra click for tables, I can live with it.

To illustrate reports, to make a case for a new outreach program / initiative.

...

Yes, downloading data for making my own graphs. This page would not support it unless I can download data (or go to if available in tabular form) from it.