In searching through Data Science job postings, they almost exclusively mention either Python or R. Is there any value in trying to keep them both active languages in my skillset? Is anyone trying to learn (or already knows) Deep Learning in R?
What are you interested in, data science or deep learning? Data science is mostly about collecting data, cleaning it up, analyzing it, making models, etc. R is a great tool for this.
But deep learning is a very specific set of techniques and is only a subset of the field of data science (or perhaps a different field altogether that somewhat overlaps with data science). Most deep learning seems to happen in Python.
Kind of both. I took a class on Neural Networks in grad school and I really like what’s happened in the years since. I was just curious.
If you wanted to learn R, might be good to focus on statistical techniques etc, since that is what it is best known for. And this would be a complementary skill set to DeepLearning in Python.
Data Science focuses a lot more on parsing data, cleaning it, and asking questions about the data (how was the data sampled? which features are important? are the features independent? how are they clustered etc). Statistical modeling techniques and R seem to be a natural fit.
Deep learning covers a specific subset of ML techniques that have been very effective in solving perceptual problems (speech, language, vision) and other problems that were previously considered to be too difficult for computers (chess, go, self-driving). The ‘inputs’ and ‘outputs’ are usually better defined here and the focus is on computation.
@bhollan, my experience is mainly R. I consider that tools are just that, tools. In R deep learning frameworks and options were less developed until recently (you were limited mostly to Mxnet, excellent but not that well documented and under heavy development). This is changing fast and soon it will just be an option to do DL in R or Python.
Anyway… to your question, is it possible to do DL in R? Sure. Does it make sense to learn both R an Python for DL? Or, better, does it matter what tool you use for deep learning? I dont think it matters that much. They are just tools and way more important is to understand the roots and ideas of what you are doing.
A frequently heard advice, its better to first be proficient in one language before learning another, and I would personaly would go first for the easiest, clearest path, that for Deep Learning is still Python. (And this advice I give you in spite of being an R lover! ).
As you have now this golden opportunity to do DL in Python that this course means I would not overthink it and would begin with Python, R is not going anywere, you can always include it in your toolbox in the future.
You don’t have to choose.
It’s pretty easy to send Numpy arrays and dataframes from one to the other.