Numpy

>also python is good because it's a general purpose programming language
which is why people in data science have to use the conda package manager with it, because pip is shit

> Use Scala

Attached: scala.jpg (400x426, 24K)

Julia

Never use R. It's a garbage language, full of bloat, retarded design decisions, bad tools (e.g. Rstudio is absolute garbage), significantly worse performance than Python with numpy, and inconsistent naming even within the standard library (e.g. function.name vs function_name). Literally the only good thing about R is ggplot2 library, which is leagues better than popular Python libraries (with ports like plotnine being very buggy).
Anyone who recommends R over Python either has baby duck syndrome, is a pretentious fuck, or can't program for shit.

Tbh I'm more likely to go with numpy because it's integrated within python so I can also benefit from other libraries like matplotlib.
The one thing I don't like so much is array broadcasting. I know this is a "feature" but this feels so convoluted I can't wrap my head around this when the shapes are 3 or more indices deep.
Unfortunately, going around this means missing out on vectorisation

>is a pretentious fuck
That's one reason I wanted to go full R for a toy project.
R feels more "ipython notebook" rather than "scripting".
The emacs package ESS also worked on my desktop but I couldn't figure out how to get it work on my laptop. So I quit R because of sour grape syndrome.

I learned python and then R
R's magrittr (%>%) and monadic FP is much better than Python's OOP hell
It makes all the shortcomings you mention completely irrelevant.
Thankfully Julia has it too, as a builtin instead of as a library (|>)

Data collection and processing in Python

Data Graphing/Displaying in R

this, except
>don't really know anyone who uses julia seriously, it's all about python and fortran in scientific computing it feels
is retarded.
I'm in physics, most profs/students are switching to Julia from MATLAB, a couple of old as balls profs use fortran but it isn't taught. Nobody i know (except myself) uses python for computing, maybe a little for some data acquisition/wrangling.
I use python and (reluctantly) R because i expect to go work in industry after i finish master's, but if i was staying in academia i would pick up Julia for sure.

isn't Anaconda just an environment manager for bundling python with compatible libraries?

en.wikipedia.org/wiki/Anaconda_(Python_distribution)

So it is just vanilla python with a package manager for people who can't into pip

yep, thats about it

it also comes with the conda virtual environment manager, which is better than virtualenv because it doesn't stuff your project folder full of binaries

numpy
use it with autojit and it does wonders
>Octave
why will you even consider this? No one uses it apart from andrew NG followers
>Julia
GL with lack of docs and bloated code

>dont use vanilla
>the Anaconda
are you really this retard? anaconda and jupyter notebooks are just a wrapper around python's shell

This is actually very true. I learnt it the hard way. Python is the best. And with seaborn, you can get pretty good plots. (not as good as ggplot2 though but way better than default matplotlib)

Is Julia retired from JAV?

Julia or Maxima

I don't use anaconda but it can't be worse than pip, can it?
I'm using pacan / AUR / manual install after pip broke everything once.

>jupyter notebooks
these notebooks are neato tho.
I'd use org mode if I knew how to display plots inline however.

Physics guy who uses python.
I have become a terrible programmer due to my python habits. This shit is like Ebonics for programming.

Why is pip shit?

Not him, and maybe my fault for being careless, but I have had a bad experience with pip.
I am the only user of my machine so I install everything globally instead of doing "one virtualenv per project" or whatever the recommendations are.
Anyway, after bloating my install with a fuckton of libraries, I had two different libraries that had a common library as requirements, but with different versions.
This fucked everything up

Scala

If stats R.

Notice how the python evangelists lie and shittalk everything not python. They have zero experience with anything.

-uses shitty OS package-manger, dependencies not there
-accidentally forget to use --user when pip installing, break system
-still have to manually tweak environment to use virtualenv correctly or use tools like pipenv.
-setup.py bullshit for packaging and distribution.
-dependencies are recorded in something that is as stupid as setup.py and something as completely useless as requirements.txt

C

Julia starts counting at 1, so numpy

Common lisp.