Update "Who Uses Chainladder-Python" docs page

[genedan]

Over the past few months, people have demonstrated impressive use cases of chainladder to the CAS Open Source Working Group. The docs have a page called Who Uses Chainladder-Python for the purpose of listing companies (or people? projects?) that use the package.

I was thinking about having a gallery-style cards layout for personal projects, or corporate open-source projects, with a screenshot/logo as the thumbnail and a brief description of what the project does. Clicking on the card will navigate the user to the project's website. Since business use cases tend to be proprietary, we could have a section that displays logos. To see what I mean, check out ratatui's homepage which has a banner with a list of companies, and projects listed below. I'm open to alternative ideas on layouts.

Since companies have rules concerning logo use, we should get their permission in writing if we are to include logos.

I think we should also reach out to those who have presented so far, if they want their project showcased on the page, either as a personal or business case.

[kennethshsu]

All great ideas, I like them!

I also think we can do a better job understanding our users too. I was thinking about getting a survey out to figure out where the users are from, and what they use it for, I think this will also help guide the project forward.

Does this look right to you? On PyPI we have 45k downloads last month? And on conda only 600 in the last 6 months.

[genedan]

I think we'd have to put together data from a variety of sources, all of which are noisy, to make a conclusion. The demos tell me that more people than I had expected use the package. I was worried that it was so niche that we were just in our own little bubble, but people do use it. I ran into coworkers who use it too.

Does this look right to you? On PyPI we have 45k downloads last month? And on conda only 600 in the last 6 months.

As far as solid numbers go, I have no idea. I look at stats on my personal website and I think I get way more page views from bots and indexers than I do from real people. If the docs are ever self-hosted, we could install things like stats counters but we'd have to be mindful of stuff like GDPR, my preference is to do everything we can to avoid needing a cookie popup, I'd rather resign myself to using noisy data than trying to deal with that.

We could take stuff from:

• Github stars
• Repo views
• Docs stats?
• Survey?
• Paper citations

And if those numbers are going up, that would be a good sign.

[kennethshsu]

Definitely.

What kinda survey are you talking about? Or do you have anything in mind?

[genedan]

I put it on the list because it was your idea! Maybe we could it include it as a question in the next technology survey.

[kennethshsu]

Ahh ok, I thought you got something else haha..

I'm not sure who will run or when the next technology survey will come. But we can ask.

[genedan]

Does this look right to you? On PyPI we have 45k downloads last month? And on conda only 600 in the last 6 months.

@kennethshsu, I've noticed the badges can have some really wonky stats. Today I only see PyPI displaying a number with no context (429). Sometimes I do see a downloads number of about 45K per month. If we want to believe the stats on pypistats, this would be closer to the correct number.

I see a much lower monthly number (hundreds) on conda forge, but with a high cumulative number of around 63K. Maybe the decrease is due to the project moving away from conda and towards uv? The charts don't go back that far in time so it's hard to verify.

[kennethshsu]

My gut is still telling me that the number is too high? I'm guessing it is inflated by indexers or web crawlers. It doesn't really make sense that there are hundreds of downloads a month... Who are these people!?

[genedan]

I think the numbers have merit, you can compare the figures to a project of mine that gets far fewer downloads.

https://pypistats.org/packages/tmval

Might not mean it's 45k people, I think it's part of a lot of CI/CD workflows and also whenever someones makes a venv. The cyclical nature of the stats is interesting, a Google search tells me it's due to people working less on the weekends.