You would have to give more information. If you need real flexibility, you would go for R and Python. If the data set are huge, so you may run into some problems with R, so Python would be a better approach. You could even run it on google infrastructure. By choose the software language, the packages of choices will be limited for a very short list. R has a very important advantage, besides of being a free available software. It has a very high standard for documentation. So, you may not only find a variety of implementations, you would find it very well documented.