I added a scope document to the project <https://g...

# mathematics

[thomasnield]

I added a scope document to the project https://github.com/thomasnield/kotlin-statistics/blob/master/version_2_scope.md

[altavir]

I think it is a good practice to keep everything modularized. Maybe move ml to submodule? I can change build to accommodate it.

[Thomas Legrand]

I’m interested to know, why is ml part of the lib ?

[altavir]

Because it is @thomasnield's bread and butter?

[thomasnield]

@altavir @Thomas Legrand I've been questioning whether ML is in scope. It can be helpful to have toNaiveBayesClassifier() and other machine learning extension functions, but an argument can be made they belong to a separate library.

[thomasnield]

I might include just Bayesian functionality (including Naive Bayes) since Bayes is certainly statistics.

[altavir]

Since I also work a lot with Bayesian statistics (not ML though) maybe it is good idea to create separate package. We currently work with those guys: https://github.com/bat/BAT.jl . I plan to add kotlin front to it.

[thomasnield]

@altavir Okay... maybe I'll slash the Bayes stuff then.

[altavir]

@thomasnield the problem is that we have sad examples of uber-packages like commons-math (they are modularizing everything now). It is good to have a lot of features, but it is better to have them as fine-grained as possible. It is quite easy to create a multi-module gradle build and move all secondary functionality to additional module. I can help with that.

[thomasnield]

@altavir agreed and thank you. Kotlin-Statistics was initially conceived as an experiment to see what stuck using Kotlin for analytics. I think it became quite clear what it's scope is now: extension functions for grouping and binning as well as statistical analysis.