this post was submitted on 12 Jun 2023
10 points (100.0% liked)

R Programming

267 readers
1 users here now

Please use this as a forum to discuss R, and learn more about it. If you have any questions about how to do specific things in R, this is the place to ask.

Getting Started

You can download R here.

You can download RStudio here. RStudio IDE, which is supported by Posit PBC, is a powerful and well-developed IDE for R. Other development environment options include Emacs addon Emacs Speak Statistics and VSCode.

Other Communities

Other communities that may be of interest across the fediverse:

Please send @a_statistician a message to recommend additional communities to add to this list.

Learning resources:

founded 1 year ago
MODERATORS
 

In the interests of making this community home for those of us who are reddit refugees, let's go ahead and introduce ourselves.

Some suggested things to comment on/include in your introduction:

  • Tidyverse, base, or data.table?
  • Are you primarily a user, a developer, or in between?
  • How long have you been using R?
  • What other languages do you use?
  • What do you use R for? Statistics? generative art? data wrangling?
  • Are you using R primarily for work, fun, hobbies, or something else?
  • Are you a hex sticker collector? Why or why not?
  • Where are you on the data engineering <----> pure statistics continuum?
  • What's your favorite obscure package?
you are viewing a single comment's thread
view the rest of the comments
[–] [email protected] 4 points 1 year ago (1 children)

Hi, I'm a PhD student in software engineering and I've used R for prototyping/testing algorithms and methods related to machine learning/data complexity/statistics. I use Python and C (for hardware related stuff). My favorite obscure package is ECoL.

[–] a_statistician 2 points 1 year ago (1 children)

Interesting! I've never seen something that tried to quantify the data complexity in quite that way before, but it looks cool!

[–] [email protected] 2 points 1 year ago

Well I needed a way to measure changes in separability in high dimensional spaces (besides most classifier performance bc that has too much variance and is not sensitive enough) and I thought this was actually kind of a solved problem but nope. So I went down the data complexity rabbit hole