Simple Statistics, a JS library for doing statistics

[shashashasha]

http://macwright.org/simple-statistics/

[noelwelsh]

The documentation for the classifier says:

"Create a bayesian classifier."

I think it should say

"Create a naive bayes classifier."

A "bayesian" classifier is a class of classifiers, not a particular algorithm. Naive bayes isn't (in most implementations) Bayesian! Finally you should specify what smoothing, if any, you use.

Hope that helps!

[dmeagor]

This looks great. For large datasets does it handle the long running request issues that cause browsers to display a warning? If anyone uses this for processing a decent size amount of data I'd be interested in your opinion.

[thelamest]

It's a very nice idea, if you have time, you could aim to incorporate other modern tools: robust statistics (trimmed means etc.), effect sizes, confidence intervals

[danmaz74]

This is interesting, but for the linear regressions I'd much rather get the a and b parameters than a function to interpolate/extrapolate y values.

[tmcw → danmaz74]

Sounds doable! Created https://github.com/tmcw/simple-statistics/issues/33 - I'll implement tomorrow.

[danmaz74 → tmcw]

thanks :)

[randall]

Recommendation: have the demos use web workers if they're available, so you don't tie up the main thread when the numbers get really large.

[bsenftner]

Thank you. Great time saver for us all coding up these! Thanks again!

[omnisci]

This may be really helpful :) Thanks a bunch! Keith