Hi Lars,
thanks a lot for your answer.
I have been working with PAL already and am impressed by its performance on k-means and some of the classification algorithms. However, the number of algorithms is very limited, and usual statistics / data mining tasks include cumbersome cleansing and pre-processing.
So, what is the best and most efficient way to do semi-complex calculations with data in HANA? As I understand it, there are the following options:
1. Use R: That will require data transfer over a TCP connection every time something has to be calculated.
2. Use SQLScript: Here data manipulations are obviously possible, but the language is somehow not sufficient more complex calculations. (How performant is SQLScript when compared to the more tightly integrated HANA functions such as analytic views? Is there also a comparable speedup doing this SQL scripting in-memory when compared to, say, PL/SQL on Oracle?)
3. Could this actually be done somehow in XS ("native")? As I understand it, this is a web/application server running on the same machine as the index server. So, can powerful calculations be done here, or is everything written in XS strictly the presentation layer side?
Thanks in advance,
Georg