I think for people who actually like to code, even a little bit, Julia is the best option. If all you want to do is run reg y x and make pretty plots, Stata or R work just fine and are likely still better right now. That might change with time as more man-hours of development get allotted to Julia.
Will academia adopt Julia?
-
instead of Julia, try K9 or Q/kdb+, the secret lang of real quants. commercial quality code w/integrated dbase in < 10 loc, & C performance because they compile to C. You'll abandon loops, iterators & procedural coding & learn function compositions -think Mathematica v Matlab
Q/kdb+ https://code.kx.com/q
K9: https://estradajke.github.io/k9-simples/k9/Introduction.html -
Yes. It will start with the DL folks and percolate from there.
Jeremy Howard, the creator of the very popular fastai python libraries has recommended moving to Julia. Yann LeCunn also says python is a dead-end for ML
https://wandb.ai/site/podcast/jeremy-howard
https://twitter.com/jeremyphoward/status/1097799892167122944?s=20 -
It's a new toy, which is cool but not mature, and there are still many fundamental problems unsolved. Matlab with mex using C++ may be better in terms of performance and robustness.
Matlab with MEX is not robust at all lol
Matlab and mex are used by many multinational corporations. They are production ready. Julia is a pure academic toy language.
-
It's a new toy, which is cool but not mature, and there are still many fundamental problems unsolved. Matlab with mex using C++ may be better in terms of performance and robustness.
Matlab with MEX is not robust at all lol
Matlab and mex are used by many multinational corporations. They are production ready. Julia is a pure academic toy language.
how is Matlab + MEX better than Python + Cython or R+ Rcpp?