Add tracks in R and Julia

[bkmgit]

It may be helpful to add tracks in R and Julia. Many of the algorithms in scikit-learn are also in R and Julia. The focus should perhaps be less on scikit-learn and more on concepts used. If one only introduces receipes, then many of the receipes may be used in inappropriate situations. Introducing concepts and showing that they can be built into different workflows may encourage better long term use.

[colinsauze]

I think R and Julia tracks could be useful, but lets try and get the material a bit more stable in Python first or else it might be a bit of a maintenance nightmare.
I feel that people should leave a Carpentries workshop being able to apply what they've learned and have access to a set of real working examples they can adapt to their use. If we get too abstract and conceptual we'll lose this. Is there any particular part of the lesson where you feel that the concept is lost because we spend too long looking at the specifics of scikit-learn?

[bkmgit]

scikit-learn does not always have the most update algorithms, for example umap can sometimes do better dimension reduction than t-sne, but perhaps it is ok to stay with just scikit-learn. So far, Carpentries has language specific lessons, so probably it is fine to do this for machine learning and have separate R and Julia lessons.

If of interest, can add a popular astronomy track based on:

• https://github.com/bkmgit/machine-learning-novice/blob/bkmgit-regression-escape/_episodes/02-regression.md
• https://github.com/bkmgit/machine-learning-novice/blob/bkmgit-dimensionality-reduction/_episodes/05-dimensionality-reduction.md