I discovered this yesterday and spent some time with it. Sadly, it’s a pretty thin wrapper, with no typing or IDE support. It’s more or less ‘evall()’ with helpful type conversion between languages.
Have any data scientists moved from R or Python to Wolfram?
Curious to know if it can do anything better?
I'd consider checking it out, if only to dabble and learn more about it. (from a ~30 second search online, unlike python and R which are free, I think wolfram is about $439/year. source: https://www.wolfram.com/mathematica/pricing/home-hobby/)
Wolfram Mathematica is probably more expensive than that, but is pretty awesome in some ways and very consistent.
Being able to build a graph network with just a single line of code where the vertices matched an image I dropped in was pretty cool and just one of a bazillion things it can do outside of just your calculus homework.
The annoying side is that it costs money, is proprietary, and all that goes with it. I can sometimes do something in a single line of code or two that would take a lot more code in Python, but sometimes it's still just faster to do in Python as the Wolfram language is some kind of Term Rewriting System that is more like Lisp or Functional programming than imperative and sometimes I don't have the right mental model and it's hard to find answers on Stack Overflow.
I think a lot of math academics love it. People use it in many fields too like engineering, economics...etc, but probably a bit niche. I assume most data science people are mostly doing Python now with a core group of stats folks still doing R or SAS or Minitab.
I cannot speak to data science / statistics but can say that Wolfram Mathematica is vastly, vastly superior to other options in non-statistical arenas of applied math and physics.
I'd definitely encourage giving it a quick try just because you are working in a closely related field and so might occasionally run into cases where Wolfram is actually a better tool for the job.
There are some very respected people that the company employs doing good work pushing the limits of different equation solvers and that sort of thing. Whatever one thinks of the man Wolfram himself, I think its an impressive company and probably a good example of a piece of software that is on the cutting edge and also wouldn't be viable as an open source project (too many PhDs required to support it for that tbh, and definitely not very accessible to contribute to).
And yes, it is expensive, but not prohibitively so for interested hobbyists. I pay for a personal edition, but my work pays quite a bit more than that. If you are affiliated with a school, see if they have deals or computers with it they can make available. Most schools with big physics or engineering or applied math programs are going to have it licensed.
It depends what you mean by "data science", but for stats, R or Origin (also expensive) would be better. Matlab (another paid option) is typically used for for numerical work (e.g. signal analysis). Mathematica is best at symbolic stuff.
This is amazing, all it needs is a few design touchups and it could even be better than Mathematica. Wish programmers had a deeper intuition there sometimes. But then again how much time would be left for the programming side.
I recommend reading a few Tufte books, make the plots a bit more fancy, tweak button positioning, styling, remove redundant notifications like when you X something and get 10 "removed" etc
As someone who occasionally dabbles in Mathematica for creative coding, using WLJS finally made me enjoy writing Wolfram code.
Their free "cloud" is so slow I'd find writing BASIC on a C64 more enjoyable while their desktop offering costs so much money that it crosses into the territory of hardware budgets.
Aside from the occasional missing widgets and bugs, it's fast, modern, and entirely local. If anyone wants for any reason wants/needs to use Mathematica, I'd recommend this as the least bad way to do so.
15 comments
[ 3.1 ms ] story [ 39.9 ms ] threadCurious to know if it can do anything better?
I'd consider checking it out, if only to dabble and learn more about it. (from a ~30 second search online, unlike python and R which are free, I think wolfram is about $439/year. source: https://www.wolfram.com/mathematica/pricing/home-hobby/)
Being able to build a graph network with just a single line of code where the vertices matched an image I dropped in was pretty cool and just one of a bazillion things it can do outside of just your calculus homework.
The annoying side is that it costs money, is proprietary, and all that goes with it. I can sometimes do something in a single line of code or two that would take a lot more code in Python, but sometimes it's still just faster to do in Python as the Wolfram language is some kind of Term Rewriting System that is more like Lisp or Functional programming than imperative and sometimes I don't have the right mental model and it's hard to find answers on Stack Overflow.
I think a lot of math academics love it. People use it in many fields too like engineering, economics...etc, but probably a bit niche. I assume most data science people are mostly doing Python now with a core group of stats folks still doing R or SAS or Minitab.
I'd definitely encourage giving it a quick try just because you are working in a closely related field and so might occasionally run into cases where Wolfram is actually a better tool for the job.
There are some very respected people that the company employs doing good work pushing the limits of different equation solvers and that sort of thing. Whatever one thinks of the man Wolfram himself, I think its an impressive company and probably a good example of a piece of software that is on the cutting edge and also wouldn't be viable as an open source project (too many PhDs required to support it for that tbh, and definitely not very accessible to contribute to).
And yes, it is expensive, but not prohibitively so for interested hobbyists. I pay for a personal edition, but my work pays quite a bit more than that. If you are affiliated with a school, see if they have deals or computers with it they can make available. Most schools with big physics or engineering or applied math programs are going to have it licensed.
It depends what you mean by "data science", but for stats, R or Origin (also expensive) would be better. Matlab (another paid option) is typically used for for numerical work (e.g. signal analysis). Mathematica is best at symbolic stuff.
I recommend reading a few Tufte books, make the plots a bit more fancy, tweak button positioning, styling, remove redundant notifications like when you X something and get 10 "removed" etc
Their free "cloud" is so slow I'd find writing BASIC on a C64 more enjoyable while their desktop offering costs so much money that it crosses into the territory of hardware budgets.
Aside from the occasional missing widgets and bugs, it's fast, modern, and entirely local. If anyone wants for any reason wants/needs to use Mathematica, I'd recommend this as the least bad way to do so.
Now we just need a FOSS reimplementation of wolframscript that can also run in the browser. Unfortunately, Mathics' one[1] is written in Python.
[1]: https://github.com/Mathics3/mathics-core
> No active Kernel is attached
While running wolframscript from the shell does work.
Do I need a Mathematica activation key?