Hello /g/, what are your opinions and toughts on R? I'm

>>56962942
R is the most used language by statisticians and so-called data scientists. Don't listen to 4chan, anon.

>>56962942
>there are better options online
Bull fucking shit.

You might use python with fast C libs instead, but R is the golden standard of statistics.

>>56962942
>1234 + 782
>Not using Common Lisp for everything
I can confirm that you can use it for everything these days. You are better off afterwards because Common Lisp is available everywhere and in every form.

>>56962971
Do not listen to him. He hasn't enjoyed Lisp yet.

>>56963010
>vector operations
>in lisp
Muh mutability

>>56963010
>obligatory CL witness
Stop shilling your religion for everything.

File: lisp_the_best.jpg (48KB, 720x544px) Image search: [Google]

48KB, 720x544px

>>56963019
I... can't... resist. As I have enough reasons to shill for it, I just can't stop.

>>56963099
Good specialist languages are better than good generalist languages in their niche.
Now unless you're arguing that R is bad, you just can't say that however awesome CL is, it's better for statistical analysis.

Language agnosticism stronk!

>>56963117
I am absolutely not arguing that R is bad at all. From what I know it is great for its purpose. But I guess it's not that great to create a GUI application from your R program/library.

That's why I argue that CL is better due it's flexibility. Other than C/C++, Fortran or Java, CL can integrate the niche specifications pretty good. You just don't feel uncomfortable when using it in a specific field as you have everything you know PLUS the specifics.

>>56962942
>as there are better options online.
I'm gonna call bullshit on that remark unless by 'online' you mean 'I can download better programs'.

Look into MATLAB and alternatively Octave as well.

File: Screenshot from 2016-10-07 07-12-09.png (62KB, 852x387px) Image search: [Google]

62KB, 852x387px

>>56962942
>>56963365
Also, if you're looking for a MATLAB alternative I've been using Julia and it's pretty nice.

>>56962942
It's still very much the"thing" to use for real science, still updated regularly, still the platform for new libraries.

Head to /sci. They'll know more about it.

People outside of the sciences like to think python and all of its little headline grabbing libraries are taking over real science programming. But it's not. Not even close.

>>56962942
In my opinion it's a great statistics tool, but a horrible programming language

>>56962971
>>56962982

thanks

>>56963010

I don't have the knowledge to argue with you like >>56963016 and >>56963019. but if there are many languages, it must be that none of them is just THE language, that serves for anything.

>>56963802
I guess it's just time until you get it enough.

>>56962942
If you're doing statistics, R is way easier to work with. Basic stats stuff is built in and you can find libraries for pretty obscure stuff easily. With python you would have to download and learn a ridiculous number of libraries to do basic stats stuff (namely pandas, scipy, numpy, matplotlib or whatever graphing library). With R, you can just dive into it plus it has ggplot2 which btfos all other plotting libraries regardless of language.

>>56962942
If you're dipping into the machine learning side of statistics then matlab and python could be good alternatives, but otherwise R is a solid choice

>>56963010
>unironically recommending esoteric programming languages for serious stuff

>>56965042
>but if there are many languages, it must be that none of them is just THE language, that serves for anything.
Ok, I am serious now: It is true that every language has its specific purpose. You pick the language like picking the tool that fits the job. But (!) many languages do not fit the job because they can't morph themselves to that domain. Lisp is so fucking general or even meta that it can be literally anything you want. For using it like R you there is: https://github.com/blindglobe/cl-blapack (lapack bindings) and https://common-lisp.net/project/cl-plplot/ (plotting) or https://common-lisp.net/project/clnuplot/ (if you like GNU plot).

I - for instance - use it successfully for my website. But I could use it equally good for a 3D game or even use it alongside Java (ABCL compiler) and contain the flexibility to still create a web service fairly easy in another part of the same program.

In the end you make the choice. So if you want to stick to the mainstream and take the advantage of it, go ahead and use R. If you want to use one tool for it all (which is possible!) and never turn back to anything else, use Common Lisp.

>>56965758
You say mainstream here like it's a bad thing: there is a reason R is popular for statistics.

Why make life difficult if you're not intending on building a career out of being a lisp programmer?

>>56962942
R is the go to free"dom" standard language for stats.
From my view it's the best to learn because it is universally recognized as opposed to other proprietary stats packages and tools. There are also people constantly contributing packages so it's hard not to find a community that shares your job function/interest and is implementing it in R.

t. Meme DS Degree

>>56962942
That logo is disgusting, 90s, ps1 tier trash.

>>56968236
Function over form sweetie. For those of us who do actual work on our computers and not just spend all day ricing the appearance of an icon means fuckall.

>>56968279
>Implying they're mutually exclusive

>>56968306

They are. Time lost is time never regained. KYS

>>56962942
If you want to do just math Jupyter + R is probably a good fit

>>56968341
Then why have a logo at all?

>>56962942
R has weird quirks, but it's better than pretty much every other statistical analysis software, paid or free, out there.

Basically, R is what you're stuck with, and be thankful your employer/school/whatever doesn't force you to use SPSS.

>>56968358
To filter out idiots who care about superfluous shit like yourself.

>>56962942

We are forced to use it (at times at least) because there are so many, admittedly well written, "standard" packages available only in R.

But make no mistake, the language is horrible, especially in terms of debugging and reliability. We try to avoid it as much as we can by wrapping routines in R using rpy2 and the like.

>>56968389
I'm glad you agree it's super shit. I'm just saying, if you're going to"loose time" making a logo, might as well use that lost time making it halfway good.

>>56968419
>making
>implying that's not automatically generated by a script

I'm writing about names, binding, and scopes for programming concepts class, one of my languages for research is R. Whats the deal with it? Why does it seem like variables have to be assigned to something just to be defined?

>>56968456
Kek

>>56962942
>but a kind anon told me yesterday that it seems outdated
the GUI might look outdated but its constantly developed, new libraries all the time

statistics stuff doesn't need the fancy 2016 bling bling flat design emojis to work

it's a powerful tool if you know how to use it, I never really got into it, my statistics courses in uni were all about R and one of my professors is constantly contributes packages

>>56968380
they did anon, that's why I'm thinking on learn R. why do they teach me SPSS instead of R? teachers said it was dificult. fuck, SPSS isn't simple neither

>>56968962
As someone who was taught both extended basics of SPSS and R.
SPSS has a lower barrier to entry, a lot of stuff can still be via the GUI and it's quite similar to excel visually.
With R you're immediately working with CLI and actual code.
If the majority of your students (think the majority of the majors that use statistics, but don't specialize in them) have never touched a CLI or computer code in their lives, immediately jumping into R can be quite confusing and scary.

>>56970409
>can be quite confusing and scary.

I agree. but my opinion is: fuck them.
normies don't feel scared or confused when they post retarded shit on internet, then they don't know nothing about computers because "it's confusing and scary" and I want to be spoonfeed when my magic box of games and porn breaks.

R is a language that is designed to do nothing but handle nested arrays (which are what table structures are in statistics). Its slow, it cant scale to large data sets. People who use R professionally constantly have to re-write R code in a compiled language that will handle large data. Anyone who uses R instead of Julia is an idiot. Julia is very fast, scales huge, can be used to write server infrastructure as well as handle data because of powerful I/O.

>>56962942
try fortran

>>56970814
should I go with julia then? why would people stick with R if it's that bad?

>>56970834
Please elaborate

>>56970917
because a lot of statistics books and machine learning books use R, just understand that R will limit you and dump it once you get a handle on what youre doing

R is awesome for statistics. I don't have that much programming experience (1 class in C++, fucking around in Lua on the side, and 40% through an R programming class right now) and can't comment on its quality for programming, but it's great for stats.

I don't know if you'll find a better program for statistics than R.

RStudio is good but crashes a bit. Otherwise it's a nice IDE for R.

>>56965758
If you are seriously recommending lapack bindings as an alternative to using R, then you really don't know what you are talking about.

>>56970917
I don't think you should go to Julia if you are just starting out with statistics. I use Julia all the time and for a beginner, R is just way easier to use. Use Julia when you're getting into serious big data stuff and you know what you're doing.

>>56971823
usually it's R for stats (vector man and specialized packages) coupled with Python for further implementation

Those two are a great match imo

>>56962942
R is pretty great and a standard tool for stats. However, it has some pretty severe limitations as a programming language. Chief among them is that it is comparatively very slow. There are also a lot of weird syntax quirks. And then there's the typing, where there are all sorts of types and classes and an object can get automatically coerced from one to another and fuck your shit up without warning.

And if you learn it as your first language (like me) you run the risk of getting stuck on R island and convincing yourself that it's a good language for everything you need so you can avoid learning other languages (like me).

On the plus side, absolutely anything you want to do stats-wise is either in the base package or in some other package.

Watch this:
https://www.youtube.com/watch?v=6S9r_YbqHy8

>>56974900
Well it's slow but only when you're trying to do non-vector operations like you'd do with C.

As someone who actually works in the stats field and isn't a neck beard, R is primarily academic and a few professional fields use it. SPSS and SAS are far more prevalent.

I watched a colleague spend almost a year fighting with IBM and R nerds trying to achieve a reliable solution to a big data problem. When he spoke to SAS, it was done in a day.

File: 1472690569058.png (289KB, 800x600px) Image search: [Google]

289KB, 800x600px

>>56963230
OP wants to do math

And for some reason you suggest he builds a gui application