Use core.matrix.stats?

[mikera]

Hi,

Just found this project - looks great! If it is maintained, can I suggest adopting some of the functions in core.matrix.stats? This includes slightly more general functions for calculating variances etc. an d I think makes a good holding place for some of the more "basic" statistical functions (without having any opinion on data structures - it uses core.matrix API functions only so is basically independent of underlying representation etc.)

https://github.com/clojure-numerics/core.matrix.stats

[ifesdjeen]

Hi, it is maintained, although realistically I'll only get to do more after July 4 :)

I'll put it on my upcoming todo, will try doing it asap.

Thanks for the hint!

[mikera]

I guess I need to know more about the library and its goals. I'm working to build the ecosystem of Clojure-related tools for numerics and data science, would be helpful I think to be clear where statistiker is supported to fit. Roughly I see it as:

1. core.matrix = fundamental API for array programming
2. Clatrix / vectorz-clj etc = core.matrix compatible array implementations (matrix, vector maths etc.)
3. core.matrix.stats = basic statistical library functions. pretty stable. Build on 1., agnostic about 2 (which practically means sticking to APIs defined in 1 and 3)
4. Higher level libraries (nurokit? statistiker?) building on 1,2,3 as needed. Ideally agnostic about 2.
5. Toolkits (gorilla-repl, incenter) - build on all the above. can mix+match libraries
6. End user applications - build on all the above

[ifesdjeen]

I'd say 4 is the closest bet for statistiker. We also have envision which builds on top of statistiker and mostly aims for visualisation, which might be a bit closer to 5. It would be great to improve the current situation with clojure and machine learning. Currently we have a bunch of closed-sourced code. Maybe we can eventually make it available.

[mikera]

Well - happy to work with you on this stuff!

If you can open source stuff sooner rather than later that is helpful so we can kick of collaboration. In the meantime I may you a couple of PRs for statistiker. Please also consider discussing ideas / developments on the Google group:
https://groups.google.com/forum/#!forum/numerical-clojure

[ifesdjeen]

Nice stuff :) Ok, I'll try to push it all.