why?

some people do

Check NP: http://nuclearphynance.com/Show%20Post. ... Key=132379

I always found the numerics on R to be lacking compared to Matlab. My numerics I mean the optimization routines. If I ran across a failure, I would find the error messages to be less than helpful. However, the usefulness of the other components of R make a strong case.I am using R more these days but that is because I am doing a lot more hadoop work.

That's not in the question. But I prefer numpy/scipy in python to those languages.You can even call QuantLib directly.It looks like Matlab but it is really cheaper (free).

That's not in the question. But I prefer numpy/scipy in python to those languages.You can even call QuantLib directly.It looks like Matlab but it is really cheaper (free).

I personally like Matlab because of its editor. The debugger is just marvelous. Also the workspace browser is extremely useful. The syntax is also neat. Documentation is complete and well integrated into the environment especially for the last versions. The performance is hard to beat from R. Using MKL underneath, Matlab sets the standard too high for open source languages unfortunately. I have to say that I am no expert in R, but I like in R its tremendous number of available packages. In statistics, R is definitely better. In fact, I found some of the Matlab toolbox implementations are even erroneous. Matlab Excel link is also more user-friendly as the corresponding RExcel. Tough, I have to say I was very positively impressed by the last RExcel release.

R is good in small-scale prototyping or as an extremely advanced calculator.But R has its deficit in dealing with large-scale data in terms of both performance and reliability.I used R for processing GBs of microarray (genechip) data for pattern recognitions -- the programming bit is similar if not better than Matlab especially for probability-based models. Matlab is more matrix-oriented.However, R's not very well written as a high-perf numerical system -- when data table is large, in its GBs, your system memory becomes some sort of scarce resourse; after a while it's often to see R stops responding, or even the system lags (in case of Windows) significantly. It does not have an optimal strategy to read file in chunks. R is not for multithreading either. Thus the performance for a lot of pattern recog or signal processing tasks is poor. Also worth noting is that the 'language' bit of R, e.g. if/while statements, have pathetic performance as it is not the 'correct' way of programming in R. You should use vectors/data adapters wherever possible.R has a strange way of handling default/named parameters of functions.R lacks an 'official' IDE but there are a few good third-party ones. There are also a large repository of 3rd party R packages, many being very useful and specialised.Debugging in R is a pain. In my days of using R (2.3.1 I recalled), I relied on printing out values of variables. That's about all I know.