Yes. I'm glad and thankful that you ask, and the importance of it is true. I'm 
thinking quickly from the top of my head things like

  * probability functions (Poisson, Binomial, Multinomial, etc.);
  * a convenient function to nicely display a statistical summary of data 
(possibly with argument to switch between pre-made display styles);
  * an overarching library to more easily steer different plotting packages 
with the same commands (I imagine this to be hard or very demanding work, 
having to keep track of many packages once they exist);
  * VegaLite;
  * typical mathematical functions for central tendency and spread (arithmetic 
mean, geometric mean, σ, MAD, etc., which is partly taken care of, I think, in 
a base library);
  * most of the typical mathematical things/functions you'd come across;
  * convenient loading and writing of different file/data formats (CSV, TSV, 
Excel, SPSS, JSON, BSON, etc.);
  * at least one well implemented stable binary storage library generally 
accepted by other systems;
  * a good convenient overarching querying library to work with data(frames);
  * something like Voyager;
  * REPL.



Those are just a few things I quickly thought of in my answer to you.

Much more generally I would of course say that you should be able to throw 
anything at it (as if at other established systems like R+ecosystem, 
Julia+ecosystem, etc.) and it will do, and anything you would throw 
mathematically at an advanced calculator.

I would probably not switch yet, even if the ecosystem is suddenly well rounded 
for statistics (in general) or even specifically for me, because of investment, 
but it would certainly be much more on my radar, knowing that together with Nim 
it could be a very good system, but I imagine it would probably start to 
attract others. I also imagine that having good no-hassle support for different 
editors is important.

That ggplotnim library looks very promising.

Reply via email to