There's no documentation, but as far as I can tell from the code it knows some distributions and can do some plotting -- but won't analyse any actual data.
Am I wrong?
(If I'm not, I should perhaps put on Github a simple correlation/OLS regression JS library I wrote using Sylvester, http://sylvester.jcoglan.com/).
We're getting one up and running on github. Sorry it's taken so long. Right now the code is being completely revamped for community involvement and extendibility. The address is https://github.com/jstat/jstat
I think these explorations are very worthwhile. People tend to underestimate how much work it takes to implement all the distribution functions.
Looking forward to seeing where this goes.
I'm hoping to see how the lib authors:
1.) deal with model workflow (for example, how do I go from model building to prediction?);
2.) explain what data structures they hope to introduce (if all the authors did was make a strong matrix library for server-side javascript, that would be enough to make everyone's day);
3.) outsource the computational grunt work to C (or Fortran, if they're really adventurous)
The core.js file will run server-side. It has no dependencies and has been included using "this", which attaches to Node.js' modules. So to use it in a script you would just have to add the following line of code at the top of your script:
var jstat = require('path/to/core.js').jstat;
As for outsourcing the grunt work to a C library... We're going to need some community involvement with that one. It would be nice to be able to compile that directly into the server, but it's not on our roadmap right now.
The browser is quickly becoming the platform of choice for a lot of what would previously have been desktop native code.
With the recent massive improvements in both Javascript performance and screen update performance, and the general availability of the canvas tag (and SVG and libraies like raphael that work using either XML in decent browsers or VML in IE<9) for client-side drawing beyond what is easy with HTML+CSS, running just about anything including some number crunching in the browser is getting more and more practical.
The browser is very attractive if you can make your app work well in that environment, and even worth some performance drop over specific platform native code, because it is a cross-platform platform: you can target Windows, Linux, Macs, and mobile platforms all in one codebase (with perhaps a few "platform" specific variations in your display code to make reasonable use of the small screens on mobile devices while still making best use of large display areas elsewhere.
I expect to see a lot more tools that you would (as little as a year or so ago) have assumed would only work well as a client-native app or a dumb-client-with-servers-doing -the-grunt-work online app, being done in browser based code.
Part of the library is visual using jquery ui. I don't believe it would work on node unless breaking the ui layer into another library. Then would could have the crunching on nodejs and the ui on the browser.
Am I wrong?
(If I'm not, I should perhaps put on Github a simple correlation/OLS regression JS library I wrote using Sylvester, http://sylvester.jcoglan.com/).