I have a question for all you data science wannabes. I have been working with the R language for the last year now and have gotten to an intermediate level. I have done several machine learning projects that I really enjoyed.
However, I see soo many articles on Python all over, even here. Therefore, I am considering ditching R in favor of Python.
Any thoughts?
Top comments (8)
Both are quite good. And I think what you learn in one, you can kind of take to the other. I've taken what I've learnt from Tidyverse and applied them to pandas (Python).
I don't think you need to "ditch" R, but it is worth picking up Python as well, I think. If you end up building a serverless ML backend on the cloud, you aren't likely going to use R for that. But if your statsmodels (Python) is behaving funky and can't converge, you can double check with R maybe.
Hey Min,
Awesome thanks!
Would you say it is generally true that R does not really have the libraries which run serverless? This is where I feel I am left behind in this area. Do you know of any tutorials that you like/recommend that go over serverless cloud work?
BTW, how is the job market in Australia? I went to Brisbane and Adelaide some time ago for work and really loved it. Is it difficult to find work as a foreigner? ;)
All good points.
The reason why JHU taught its older Data Science MOOCs in R is really only known to J. Leek and R. Peng. As far as I know this has not been discussed. The rest is speculation. Although, I think you are correct to speculate that it comes to personal preference & biases.
I have taken several of Leeks MOOCs and it's funny to note that he does not use ggplot2 (and the tidyverse, I believe) as much base R graphics. He says that base R is much simpler, easier and more productive. That is something I knew a long time ago. It may not be pretty but it damn well gets the job done.
Back to my point(s); For me the question (Now) becomes did they think R will progress and stay extensible and keep up by itself. Will R move forward into the future easier and faster than python seems to be doing. (I say no.) This is the answer that I would like to see from Leek and Peng.
My point (now) is that R alone doesn't allow the easier transition to the web that Python seems to have the advantage in. R does not interact with the web with flexibility and robustness in the same way that python has with flask and django. In my eye, Python can now be used more directly in a larger variety of situations.
I am wondering as you just use Python in RMarkdown as well. I don't really know when you will absolutely need Jupyter Lab or CoLab.
Good point, but I don't use python in Rmarkdown all that much. I use R like a hammer. ;)
You have another good point about using CoLab. I have played with that very briefly. Is that something you use much or is it just a novelty?
I only use CoLab briefly during a short training.
I used to use Jupyter Notebook a lot, and know very well that you can transfer data between Python and Javascript.
Ya' know, I have come to the same conclusion.
Thanks for the comment, it reinforces my thinking.
TY ;))