Python vs. R

Which is better for Data Science?

Other urls found in this thread:

shiny.rstudio.com
twitter.com/NSFWRedditVideo

excel

depends. if you're gonna be making applications, python is a more general purpose language and can be used to make modules, executables, etc.

as far as I know, R can only do the data analysis part. so you'd have to couple this with some other language to build applications/deploy it. i don't know what R outputs to, maybe csv or xls files you could potentially use R by itself to generate files and then use something like C++ or Python/PHP to create applications.

you'd have to do the same with python but you have the ability to use the same language for analyzing the data and also building an application.

correct me if i'm wrong

You're right in the sense that python is more flexible and that if you want to program actual applications you should probably stick to it. BUT technically, R is just a programming language as well. If you wanted to you could program anything into it no matter what. Would be highly inefficient and fucking stupid but you could theoretically do it

R Shiny - RStudio
shiny.rstudio.com

This.

Both are often required by clients employers.

>clients employers
employers*

Python. You can run code via dask and numba that will run as machine code on as many machines as you like via llvm. PySpark works great with hdfs. R for the most part is all about fitting your data into memory on one machine.

>DATA SCIENCE

Either are good. I use Python myself, it fits better with other stuff

both.

Both are probably about to lose market share to Julia in the future, owing to Julia's superior design for computational sciences. Python is superior to Julia for many other applications, but nothing prevents R from being completely replaced by Julia.

Going to learn R this winter as part of an economics course at uni. Also hoping to be able to use it in stock market backtesting.

What should I expect? Got no programming experience, superficial familiarity with web design and python.

Tell me more

Its a new meme for people who use stat methods but dont know a ton about them because they arent good enough at stats

>Mfw i used R during my summer-job at UNI and actually have experience with it

R can do everything python does. However it only truly outperforms in the statistics department, for all other tasks python is quicker in at least dev-time.
It is truly a hell to work in though if you want to make real applications. Don't even ask me about the 6 different types of normal classes there are.

See Its hell for applications. And get used to things like this kind of list syntax.

list_data

>nothing prevents R from being completely replaced by Julia
There is CRAN. Most of the libs there have little change of ever being ported.

So? It is an index-1 language, as is awk and if I recall correctly MATLAB/Octave. `list_data[1]` returns a list actually. `list_data[[1]]` returns a numeric someone from other language will expect.

Index-1 languages stopped being a thing for a reason.

Julia is designed to be used for computational science, just like fortran was. The JIT compiler is only part of what makes it faster, the entire language is designed to make it comparable in execution speed to a compiled language. Julia is likely to mean the end of the two step approach (write computationally intensive parts in a compiled language and the rest in a high-level/scripting language) used today. R and Python cannot do this, and never will. Changing them enough to make this possible would essentially turn them to new languages with no backward compatibility.

It makes sense on low level languages but there is no particular reason for having a index from 0 on higher level ones. I looked it up and besides R and MATLAB others are Fortran, Wolfram, Julia. So technical languages have a thing for index-1.

lolno

Holy shit thanks for the insight. Is it useful to learn it if you don't have advanced knowledge of applied maths?

Not trolling, why does it makes sense in lower-level languages?
In lower level languages you mostly just use arrays, and doing array[0] just means you access the object located at mem_location_of_array + size_of_object_in_array * index, meaning 0 makes more sense.

If you ever use MATLAB, Octave, Fortran, R or Python+NumPy/Pandas, then yes. If anything it will at least partially replace MATLAB.

In higher concept languages it's easier to model math formulas and such with 1-indexed arrays.
For any other kind of programming language though zero based is obviously better. But that's the thing, programmers aren't the ones using MATLAB and Wolfram

>why does it makes sense in lower-level languages?
>meaning 0 makes more sense
I think you answered yourself there.

My bad guys just read your reply backwards.