3 ms·

How does R compare to Mathematica for data science ?

by wiz21c 4y ago

wiz21c 4y ago
Well looking at Mathematica's documentaion, the notebook-like presentation of examples is definitely top notch.
epgui 4y ago
While R is not a great language, it's a great language to work with. It has a huge ecosystem of libraries and a very active community. It's unmatched for statistical analysis (even by python), so it does great for data science. I'm sure Mathematica could be great too, but I'm not sure how big of a community it has.
fn-mote 4y ago
The ergonomics of R are terrible compared to Mathematica. (My opinion, obviously.) In R, packages are developed by separate authors. Similar functions don't have similar interfaces. There's no unity of style. In Mathematica, much more careful thought was given to everything, including at a very basic level: names of functions. Now I'm a big fan of the Tidyverse, and the fact that such a post-hoc language cleanup was possible (and done) is amazing... but the fact that it was necessary [1] is not. [1] Really not intending flame-bait. This is an opinion that many people disagree with. Perhaps like systemd. It isn't the substance of this comment, it's just the context for my remarks.
- wiz21c 4y ago
  For what I know of R (about 1.5 year full time with it; probabilities modelling/data analysis), I concur: the ecosystem is rich AND messy. I'm super impressed by the quality of everything, but it's often a "needle in a haystack" experience... Interstingly, I was a strong python pogrammer so I started to do my stuff with scipy and so on while cursing at R every 5 minutes. And after that year, tables have turned, I much prefer to work with R...
laichzeit0 4y ago
R is great if all you do involves working in RStudio and that is where it stops. The R runtime is painful to scale. You’re not going to do anything like FastAPI + scikit learn model serving. All the tooling is years behind what’s available in the Python ecosystem. I can’t speak for RStudio and their commercial offering since I’ve never used it, nor want to be locked into some vendor like that. I’m going to be honest, even RStudio sucks compared to PyCharm. Like really sucks. I almost believe people who love RStudio have never used a better IDE or don’t have serious software development backgrounds to know that it sucks.
j7ake 4y ago
R was originally intended for statisticians, while Mathematica is more for math and physics. For data science, R is probably better, just because of the wide range of common and niche statistics software.

How does R compare to Mathematica for data science ?

by wiz21c 4y ago

How does R compare to Mathematica for data science ?

wiz21c 4y ago

Well looking at Mathematica's documentaion, the notebook-like presentation of examples is definitely top notch.

epgui 4y ago

While R is not a great language, it's a great language to work with. It has a huge ecosystem of libraries and a very active community. It's unmatched for statistical analysis (even by python), so it does great for data science. I'm sure Mathematica could be great too, but I'm not sure how big of a community it has.

fn-mote 4y ago

The ergonomics of R are terrible compared to Mathematica. (My opinion, obviously.) In R, packages are developed by separate authors. Similar functions don't have similar interfaces. There's no unity of style. In Mathematica, much more careful thought was given to everything, including at a very basic level: names of functions. Now I'm a big fan of the Tidyverse, and the fact that such a post-hoc language cleanup was possible (and done) is amazing... but the fact that it was necessary [1] is not. [1] Really not intending flame-bait. This is an opinion that many people disagree with. Perhaps like systemd. It isn't the substance of this comment, it's just the context for my remarks.

wiz21c 4y ago
For what I know of R (about 1.5 year full time with it; probabilities modelling/data analysis), I concur: the ecosystem is rich AND messy. I'm super impressed by the quality of everything, but it's often a "needle in a haystack" experience... Interstingly, I was a strong python pogrammer so I started to do my stuff with scipy and so on while cursing at R every 5 minutes. And after that year, tables have turned, I much prefer to work with R...

laichzeit0 4y ago

R is great if all you do involves working in RStudio and that is where it stops. The R runtime is painful to scale. You’re not going to do anything like FastAPI + scikit learn model serving. All the tooling is years behind what’s available in the Python ecosystem. I can’t speak for RStudio and their commercial offering since I’ve never used it, nor want to be locked into some vendor like that. I’m going to be honest, even RStudio sucks compared to PyCharm. Like really sucks. I almost believe people who love RStudio have never used a better IDE or don’t have serious software development backgrounds to know that it sucks.

j7ake 4y ago

R was originally intended for statisticians, while Mathematica is more for math and physics. For data science, R is probably better, just because of the wide range of common and niche statistics software.