You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
A model that predicts 99% chance of something happening is likely inaccurate. Models tend to focus on a small number of factors & make many strong assumptions. They may be accurate as long as their assumptions are verified -- but reality is messy and full of unknown unknowns.pic.twitter.com/5Il0yB46oQ
This applies to machine learning: remember that your model's metrics are making the very strong assumptions that 1) your distribution is static, 2) your test data is representative of the real world. Your 99% accurate model won't be 99% accurate in production.
Good point by @fchollet. Even if you exclude the false 95% and 99% accuracy claims produced by poor rigor, methodology, or fraud, many algorithms fail to account for uncertainties or errors in input data. Physicists seem much better at thinking about and adjusting for this.
So what do you do when you need a model to detect anomalies that happen in 1% (or less) of real world casea?
Focus on sensitivity instead of accuracy
as certainty tends to 1, metacertainty tends to 0
Ooh, looks like somebody's subtweeting 538 
Perhaps in general, but 538 do check their accuracy after the fact, and do remarkably well: https://projects.fivethirtyeight.com/checking-our-work/ …pic.twitter.com/exfgkCil3O
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
