> Always use Python to process empirical data... Quite a sweeping statement based on a single benchmark with no code provided, and where one of the contenders

> Always use Python to process empirical data... Quite a sweeping statement based on a single benchmark with no code provided, and where one of the contenders is backed by a database. I use Mathematica for datasets many times his dataset size (350 MB) on a regular basis. Certain operations may be easier to write in other languages, but it's definitely possible to write fast and memory-efficient data processing code in Mathematica.