QuoteOriginally posted by: ZmeiGorynychI'd say learn matlab instead, or maybe R. The relative advantages of these two were discussed a lot in different threads in the software forums.SAS, SPSS, Stata, Eviews etc. may be good for statistics, but IMHO suck for everything else. Matlab (my personal preference) and R are all-purpose data processing tools, much more versatile.S-Plus is both more expensive and less powerful than either matlab or R, by a long way. Some people still use it, but everything gets used by somebody Disclaimer: my perspective is that of a high-freq quant/[trader] in a major IBI favor the Matlab/R combo as needed, but SAS pops up every so often.

QuoteOriginally posted by: ZmeiGorynychI'd say learn matlab instead, or maybe R. The relative advantages of these two were discussed a lot in different threads in the software forums.SAS, SPSS, Stata, Eviews etc. may be good for statistics, but IMHO suck for everything else. Matlab (my personal preference) and R are all-purpose data processing tools, much more versatile.S-Plus is both more expensive and less powerful than either matlab or R, by a long way. Some people still use it, but everything gets used by somebody Disclaimer: my perspective is that of a high-freq quant/[trader] in a major IBIf you can understand and use SAS pretty well, it is very easy to pick up Matlab or R. As in my previous reply, SAS is more widely used in (serious) research group while Matlab is more widely used in (quick-and-dirty) front desk analysis roles.Disclaimer: my perspective is that of a quant modeler/strategiest in a bulge bracket IB.

QuoteS-Plus is both more expensive and less powerful than either matlab or R, by a long way. Some people still use it, but everything gets used by somebodyThe S language, originally developed at Bell Labs, is generally available in two implementations, the commercial system S-Plus and the open-source system R. Over time, and because of some detailed internal differences, the two implementations have diverged in some details. I have used R a lot for my research and find that it is well suited for doing research level statistics. Indeed you could argue Matlab is a lot faster, which is true, but then you have to program up some sophisticated statistical algorithms from scratch which takes time hence R is the tools of choice in some statistical research communities and the libraries are regularly updated with the latest algorithms from top academics in the field. Matlab is more a general tool for mathematical programming. SPSS and SAS are used heavily in areas of market research, marketing, etc. by statisticians. SPSS is more a user friendly stats package and has some decent output that can readily be exported to your presentations and excel sheets. It tends to be used more by social scientists to analyse survey results, sampling issues etc..I think Matlab/R combo will probably be a good starting point to learning programming. Then you can move onto C++ and find that it takes 4 lines or more of code to do the same task in matlab/R (what a pain!)Stats Guy

QuoteOriginally posted by: LSEstatHow do you learn SAS by yourself? You have it on my laptop? I never got a chance to learn it, cannot afford to buy it myselfThe way I learned it outside of school/work was to access it through remote desktop/Citrix. I am not sure if there's any way to get around it if your employer doesn't have SAS.

QuoteOriginally posted by: csaQuoteOriginally posted by: LSEstatHow do you learn SAS by yourself? You have it on my laptop? I never got a chance to learn it, cannot afford to buy it myselfThe way I learned it outside of school/work was to access it through remote desktop/Citrix. I am not sure if there's any way to get around it if your employer doesn't have SAS.You can write SAS code in a text editor (e.g. notepad), then compile this code into Java using an open source SAS compiler called Carolina ( http://www.dullesopen.com/). You will need to learn how to compile this java code (e.g. javac mysascode.java), then run it (e.g. java mysascode).You do not need to know how to program in Java.

as a headhunter in this space I am seeing that R is becoming more and more popular with some of my clients.

Quote as a headhunter in this space I am seeing that R is becoming more and more popular with some of my clients.Looking at the quant jobs advertised more recently it seems having a PhD in Bayesian statistics coupled with R programming would make you hot property at the moment Stats Guy

sorry about the non post. This is very true StatGuy.

SAS Base is not allowed to download from thepiratebay.org either