Julia used to be infamous for its poor support and scarce libraries, AFAIK, but it seems it has improved significantly. Now it additionally enables such ambitious features as polymorphism / despatch at runtime, etc., from what I'm reading. I guess if you are only doing data analysis and don't require performance optimisation, you won't have much use of it.
Still, in the nowadays job market, I'd advise getting familiar with Python in the first place (have a look at iPython if you like Matlab environment). Matlab/Octave sound a bit amateurish or fusty. Just my impression.