StackExchange is an always-growing collection of QA sites, new sites are proposed every day at Area51 and they have to go through define, commit, private beta, public beta before becoming a full site. Many get closed during one of these steps, based on informed opinions of a number of StackExchange employees.

Let's say I want to check whether the Japanese Language QA site is likely to get closed soon.

Is there a webapp that allows me to check easily, without having to design models and crunch data by myself?


  • Predictions for all sites (any stage)
  • Good model using all past data. A single model for all sites is OK, but more detailed models are also OK.
  • Sample data is relatively limited for now (70k for proposals, 150 for betas). So indicating a confidence score for each prediction would be great.
  • Free
  • Data is refreshed regularly
  • Ways to interact a bit with the data would be welcome, but not strictly necessary. For instance overlaying the statistics histograms of 2 sites and get the "similitude ratio" for each metric.
  • A few months ago I was thinking of doing the same! For closed SE data dumps are available on Area 51. That should be a nice ML project :) Jun 19 '14 at 15:19
  • Yeah. And again something for StackApps (please forgive me mentioning that all the time, but being specifically for SE, that's where it should be listed at least ;)
    – Izzy
    Jun 19 '14 at 18:30
  • 2
    Given the small sample size, this seems grossly unrealistic. Jun 19 '14 at 20:30
  • 3
    A search for “has been closed” turns up 25 results of which 1 is a false positive. I think it's missing the ones that failed and were rebooted, but there are only a handful of those. Jun 20 '14 at 8:47
  • 1
    Even if you analyze all proposals, I don't think you'll get a useful result. I strongly suspect that you'll find a very good predictor of launching is making it into beta. Jun 20 '14 at 8:49

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.