Right now, I'm the only person reporting using R, which I am trying to learn, to substitute for my longtime usage of SAS.

Having taken Professor Ng's Machine Learning course on Coursera, I learned Octave (Matlab-like freeware), which I was using on Windows. I decided to work on this course-- Learning from Data-- on OS X and was preparing to install Octave when I decided to see which software packages are the most popular for machine learning.

**After some exploration I found this interesting chart: Kagglers Favorite Tools**

I then decided to jump into R (I have had considerable training in statistics, and was exposed to R during that training, but have been able to accomplish basically everything I ever needed to do with SAS.) Because I left myself not enough time, I had to use Octave to do the first problem set, but after the submission deadline I re-did it using R. I also used R to do the second problem set and plan to use it for the remainder of this course.

As a statistician, using R makes sense, because of the wealth of statistical tools it has. I have been trying to migrate to R from SAS for some time, and this course is a great MacGuffin to do so.

I decided to post about this decision-making in case anybody else was wondering about which software package to use and was wondering how we each chose the package we are using.

My longtime usage of SAS locked me in to that package for years, so the languages we choose to use initially often become the languages we use indefinitely. I am trying to reboot my choices by learning to use R for data analysis, Python for data processing, while only dabbling with Matlab/Octave.

However, as sets of data ever grow larger every day, I wonder which software package is (or will be) the fastest/most efficient. The amount of time a developer takes to construct an algorithm matters, but as N goes toward infinity, so does the amount of time the computer takes to implement that algorithm.