Any one doing R programming?

shawnxxiong

Currently doing a project on R, any one interested in sharing?

shawnxxiong

good open source, originally kinda limited to statisticians. but has been broaden to web developers as its available packages growing.

Amirani

I've done a fair bit of it in college. What are you looking for?

shawnxxiong

NiallSparky wrote: »

I've done a fair bit of it in college. What are you looking for?

How do you feel about R? Intuitively, I think it is a programming language specialised for a certain group of expertise, like statistisian. But when my supervisor told me he want also the visualisation part to be as interactive as ajax or flash, my life's kinda miserable. Who the hell will use R for data interaction? I think even googleVis is very limited compared to my supervisor's requirement

Amirani

shawnxxiong wrote: »

How do you feel about R? Intuitively, I think it is a programming language specialised for a certain group of expertise, like statistisian. But when my supervisor told me he want also the visualisation part to be as interactive as ajax or flash, my life's kinda miserable. Who the hell will use R for data interaction? I think even googleVis is very limited compared to my supervisor's requirement

My usage of R in college was pretty much exclusively as a tool for processing statistics. I never went much further on visuals than plot and tsdisplay.

In terms of how I feel about it; it's very useful but it's definitely a specialised sort of area.

shawnxxiong

NiallSparky wrote: »

My usage of R in college was pretty much exclusively as a tool for processing statistics. I never went much further on visuals than plot and tsdisplay.

That's what i expected, but it's totally a different story when start the project...

Sarky

I did a lot of R in my bioinformatics masters, it's really useful for mucking around with large datasets and turning them into pretty pictures. My biggest gripe is the lack of a decent user manual, and the various packages are often very poorly documented. It's bloody hard to pick up unless you have a decent teacher who's already fluent in it. Definitely easier languages to create visually interactive stuff in. But I suspect if you can get the hang of it in R then you can do some really nifty stuff.

shawnxxiong

Sarky wrote: »

I did a lot of R in my bioinformatics masters, it's really useful for mucking around with large datasets and turning them into pretty pictures. My biggest gripe is the lack of a decent user manual, and the various packages are often very poorly documented. It's bloody hard to pick up unless you have a decent teacher who's already fluent in it. Definitely easier languages to create visually interactive stuff in. But I suspect if you can get the hang of it in R then you can do some really nifty stuff.

some packages of R are just awesome, although agreed on "they are poorly documented". well, it depend on how deep you wanna to be "interactive", googleVis could be a solution (the best one TBH), but if you wanna make a cool application by R, you will need pretty much others, like html, ajax, js, php sort of things.

Dr.MickKiller

You could have a look at rpy, which is a python interface to the R language. You might be able to use python to do the web front end and then just call the bits you need in R using rpy.

shawnxxiong

Dr.MickKiller wrote: »

You could have a look at rpy, which is a python interface to the R language. You might be able to use python to do the web front end and then just call the bits you need in R using rpy.

Just inspired from your answer, what do you think of R's future? for me, it will be either a language for a certain group of people or running background like a mediator between UI and data repositories. no chance for a mainstream.

fergalr

shawnxxiong wrote: »

How do you feel about R? Intuitively, I think it is a programming language specialised for a certain group of expertise, like statistisian. But when my supervisor told me he want also the visualisation part to be as interactive as ajax or flash, my life's kinda miserable. Who the hell will use R for data interaction? I think even googleVis is very limited compared to my supervisor's requirement

I wouldn't recommend using R to write flashy interactive visualisations. (I did some work rendering openGL with R; if I had to, thats the way Id go. But I wouldnt use R, and I would tell my supervisor I was using a more appropriate tool for the job.)

shawnxxiong wrote: »

Just inspired from your answer, what do you think of R's future? for me, it will be either a language for a certain group of people or running background like a mediator between UI and data repositories. no chance for a mainstream.

I have pretty strong feelings on R.

I played with it a good bit when I started my PhD.
I also played with Matlab and a few other environments.
I know quite a few general purpose languages; my undergrad is CS.


I think R is a mess of a language, and that it is unlikely to have a very bright future.


Clearly it has fantastic libraries for statistics. In many cases, if you want to do something simple, that is pretty close to the planned use case of a library, R might be great.

But, in my only moderate experience with it, the language is very flawed and poorly designed. I don't really like the syntax or semantics. I think its not easy to plug together modules and tools written by different authors. Even after a good bit of time using it, I found I still had to look up how to do basic operations similar to other operations I did before, and I found that way I was supposed to do those operations was very surprising (which should ring all the language alarm bells).

Ultimately, I think Python is going to destroy R. Python is big in the same broad space (scientific computing, data analysis). R still has many more libraries and packages, for many types of stats work, but I think python will gradually get more and more of the more common packages. Python is also very well designed, and getting better all the time; a very tough language for R to compete against in the long game.

magpie0508

fergalr wrote: »

I think R is a mess of a language, and that it is unlikely to have a very bright future.

it doesn't seem like a programming language for me at very first glance.

fergalr wrote: »

Ultimately, I think Python is going to destroy R. Python is big in the same broad space (scientific computing, data analysis). R still has many more libraries and packages, for many types of stats work, but I think python will gradually get more and more of the more common packages. Python is also very well designed, and getting better all the time; a very tough language for R to compete against in the long game.

maybe the inventor of R meant it to be exclusive.

Manach

I've used in industry. It does seem to have an awkward syntax. But I enjoy using it to knock together quick and informative graphs for clients.

magpie0508

Manach wrote: »

I've used in industry. It does seem to have an awkward syntax. But I enjoy using it to knock together quick and informative graphs for clients.

how is it perform when dealing with big data? for me, when running the application on my own computer, it could be kinda slow especially when involve visualisation packages such as googleVis. but never used it in industry.

fergalr

Manach wrote: »

I've used in industry. It does seem to have an awkward syntax. But I enjoy using it to knock together quick and informative graphs for clients.

It can be very useful if you want to do something simple, a common use case.

I find it breaks down when you want to do something just a bit more complex, or off the beaten track.


I don't mind tools that have a specific niche that they are good at.

I don't like tools that give trouble when I try and use them to do something similar, but slightly different.

fergalr

magpie0508 wrote: »

how is it perform when dealing with big data? for me, when running the application on my own computer, it could be kinda slow especially when involve visualisation packages such as googleVis. but never used it in industry.

Poorly, but thats ok, because no one has big data.

Edit: Typed in the lull waiting for a SVM to train...

Manach

magpie0508 wrote: »

how is it perform when dealing with big data? for me, when running the application on my own computer, it could be kinda slow especially when involve visualisation packages such as googleVis. but never used it in industry.

The datasets where never too huge. Usually I aggregated various columns in the DB, exported them into a CSV and read them into R so that they would fit into monthly or else 4 week concurrent line charts - to look for anomalies.
Also used R + choropleth maps of the US - a nice way to visualise how the client system was performing. There are other country maps available, but never had need to use them

Sarky

fergalr wrote: »

Ultimately, I think Python is going to destroy R. Python is big in the same broad space (scientific computing, data analysis). R still has many more libraries and packages, for many types of stats work, but I think python will gradually get more and more of the more common packages. Python is also very well designed, and getting better all the time; a very tough language for R to compete against in the long game.

This. I've had to write a half dozen programs in Python over the last few weeks to format and analyse some large datasets, and it was almost enjoyable. Doing the same thing in R would have taken 10 times longer and had me pulling my hair out in frustration at the way the language actively tries to get in the way, while cursing the human race to the blackest depths for spawning whoever came up with the language. I'd prefer to do it the longer, more complicated way in Python than the downright esoteric way R would handle it.

AltAccount

I came across this, but it might be too basic for you OP.

http://www.youtube.com/playlist?list=PLOU2XLYxmsIK9qQfztXeybpHvru-TrqAP

SimonTemplar

R is part of my A.I. module in my next semester in college. I'm unsure why because all my research into it says it is for statistics and not A.I.

Can R also be used for A.I.?

fergalr

SimonTemplar wrote: »

R is part of my A.I. module in my next semester in college. I'm unsure why because all my research into it says it is for statistics and not A.I.

Can R also be used for A.I.?

R is a complete programming language, so, sure, you can write code in it to do pretty much anything.
But its not really aimed at being a general purpose language, so the question is, would you want to write AI code in it?

That depends somewhat on what you mean by 'AI'. If the course is mostly focusing on machine learning, R might make some sense.

There's a big intersection between modern applied AI and statistics at the moment.

Other more traditional forms of AI might be more associated with Lisp or functional programming styles.

It depends on what sort of 'AI' the course is focusing on.

Wheres the course, and whats the context, if you don't mind me asking?

SimonTemplar

fergalr wrote: »

R is a complete programming language, so, sure, you can write code in it to do pretty much anything.
But its not really aimed at being a general purpose language, so the question is, would you want to write AI code in it?

That depends somewhat on what you mean by 'AI'. If the course is mostly focusing on machine learning, R might make some sense.

There's a big intersection between modern applied AI and statistics at the moment.

Other more traditional forms of AI might be more associated with Lisp or functional programming styles.

It depends on what sort of 'AI' the course is focusing on.

Wheres the course, and whats the context, if you don't mind me asking?

Thanks for that.

The course is a BSc in Computing. I'm going into third year and the module is called "Introduction To Artificial Intelligence". A few books on R is in the reading list.

The module breakdown is

• Introduction to Artificial Intelligence (20%)
  • Foundations of AI: philosophy, maths, psychology, computing, linguistics, logic, probability theory • Historical evolution of the field • Weak vs Strong AI
• Intelligent Agents (20 %)
  • Percepts, Actions, Goals, Environment • Rational Agents • Environments • Agent functions and programs • Simple Reflex Agents • Reflex agents with state • Goal based agents • Utility based agent
• Search-Based Problem Solving (20%)
  • Utility based agents • Performance • State space search • Uninformed Search strategies • Uniform Cost, Dept-First, Depth-Limited, Iterative Deepening • Informed Search strategies • Best First, Greedy Search, A* Search • Heuristic functions • Iterative Improvement algorithms • Hill climbing & Simulated Annealing
• Game Playing (20%)
  • Hill climbing & Simulated Annealing • Games as search and Types of Games • MiniMax algorithm • Imperfect decisions • Evaluation Functions • Alpha Beta Pruning
• Logical Reasoning Agents Playing (20%)
  • Knowledge based agents • Knowledge representation • Logic • Reasoning and Inference • Propositional Logic • First Order Logic
• Teaching methodology
  The learning strategy involves the use of lectures, tutorials, case studies and class discussions as appropriate. Students will also have access to web based support.

fergalr

That's not a machine learning focused course.
That's a more traditional AI course.

I would guess any general purpose language would be fine, and don't see a particularly good reason to choose R.

But maybe there's a particular optimisation library in R that the lecturer wants to use - there could be a good reason for choosing the language.

ClaireIoana

Hi, we are doing a project using R. We were given two spreadsheets containing horse racing results and we were asked to predict the future outcomes using the spreadsheets. Any ideas or help would be greatly appreciated.
Thanks!:)

srsly78

Which part are you having trouble with? Post up your work so far.

PIKOMAN

Hi,

Anyone know where I could get some Grinds in R, I started a course on data analytics and I need some help to get my head around the R coding part of the course. Any advice greatly appreciated

Thanks