R vs Python: Which should you learn for reproducible data science (CC168)

R vs Python: this phrase has over 4.25 billion results on google. Perhaps you've read a few of those results. But how can you make sense of all those posts? Should you learn the R or Python programming languages if you want to improve the reproduciblity of your data analyses? In this episode of Code Club, Pat will break down the different considerations, caveats, and questions to ask before telling you what you should choose.

You can also find complete tutorials for learning R with the tidyverse using...

0:00 Getting started is the most important thing
2:29 Compiled vs. interpreted programming languages
4:31 What interpreted languages should we consider?
12:12 Python or R, which should you learn?
17:45 What about those blog posts you read?
19:57 The value of learning the first language

[Riffomonas]

What would make learning a programming language easier for you?

[luisandresmorenocueva]

Your content is so underated! I just discovered you and I can belive I've been learning data science without you. Keep the friking good content and sharing your knowledge.

[surfingbilly]

Heres my opinion:
- Go with R, if you are purely going to be doing statistical analysis/reports, especially if you are in specific niche fields, R has access to a lot more specific statistical packages than python so you will be able to run them with one line of code instead of having to hard code them (which can be very tricky and time consuming to do).
- Go with Python, if you need to do a mix of statistical analysis and software development, for example if you have to integrate data analysis into a backend system, or you have to include web-scraping systems, etc. Basically whenever you have to integrate statistical analysis with software even if you aren't directly responsible for the software side of things.
- Go with Python, if you are planning to also start learning machine learning/deep learning/neural networks, etc. The packages in python for this are very good (e.g. tensor flow).

[johneagle]

Exactly! I use R, i love R. I tried Python, and I do not know why I did not like it. But, I always tell people that the best language is the one they like the most. By the way, your content is really, really good. Thank you very much!!!. I am learning a lot watching your videos.

[morgomi]

I use R for machine learning, statistics, vis and for fun :)) I use Python for deep learning and image processing. Both languages are fun but I always love the R community rather than the Python community and I don't know why :))

[rickdoesai]

Great stuff! It can be daunting to start CS and be confronted with all these options, if you're self learning it's probably best to just get started and pick a high level language, then work your way towards programming 🤖

[stretch]

Great advice. One thing I wish I could tell myself in hindsight is that if you are working fairly solo going with the language that has the most tooling available for your topic/area will make your life easier when starting out. If you're in genomics/proteomics then it's hard to go past the sheer number and variety of tools available in R, if you're into serious modelling and climate science Julia seems to be excelling in that domain, etc. my 2c added to a good video!

[bavitang]

Pascal was the first programming language that I learned too! And I'm not that old, they're just slow to update the curriculum in my country 🤣. But I'm eternally thankful to my CS teacher back then because he gave me a good foundation to tackle other languages.

[antoniocastan]

Really good advice! My first language is spanish and I love speaking in english. I feel the same way about R and Python.

[AndikaMcenroe]

Thanks your video is strengthening my stance as R users among python team :)

[Daikoro]

This video was great! Would you perhaps do a video on how to read or run Python code in R? Just to demonstrate that python resources and functions are still accessible to R-users

[YannC-pq]

I love how julia is always a honour guest in these debates

[shashi_kamal_chakraborty]

Agreed! Nicely explained with apt justifications👍 Well, I decided to go with the both langs, being little biased towards R.

[bigboy]

Add me to the stack of folks who are saying thanks. My journey with programming was, very difficult. The first language I learned and LOVED was QBASIC [an off shoot of BASIC] as a kid. And I just knew I was going to be making video games.

Fast forward to college and I STRUGGLED to learn C & C++. I switch majors from computer science to photography lol.

I had every kind of warehouse & retail job. But my current job needed someone who could use excel and SQL. I was proficient in excel and new basics of SQL and zilch Python & R. Then two data analysts above me left the company and in a panic hired me, someone who had no data analysis history, to do basic daily tasks.

I didn't want to get fired or put back down into my call center position. So I youtubed and stack overflowed and two years later people are calling me an experienced data scientist. I have no idea, at the age of 40, why R is my jam. But I thank you for helping me along my journey.

[cyrillejar]

Thanks ! I was a bit afraid you told us python was better ! Before this video I was wondering if I had to learn python, whereas I worked hard to have a basic R level...

[jahidhossain]

i really like your videos.. your videos are much appreciating for me to learn R... could you please make a series of videos for biological data with R?

[niceday]

For a new beginner, get on the bus no matter R or Python, both are great! And it does not bother to learn the other one for
specific usage if it is needed.😁

[jaredwsavage]

I use both R and Python, but I tend to use R a lot more. I find it easier and more more interpretable and has some amazing libraries like the Tidyverse andTidymodels, and even the built in functions are great (And RMarkdown is just superb). The main reasons I learned Python at all is that is the most common language in the data science community, particularly on Kaggle. A lot of delevopment also goes into python first, for example, Tensorflow/Keras are developed primarily for python and while these are also available in R, it's basically a thin reticulate wrapper around the Python version.
My main problem with Python is the lack of consistency. For example, the Zen of Python states: "There should be one-- and preferably only one --obvious way to do it." Why then, is there at least four separate ways of adding layers to a Keras DNN? I know this isn't strictly Pythons fault, but it is typical of the Python ecosystem in general.

[addictedyoutub]

Learn both. Both are good complementary to each other.

[dimitrioskioroglou]

I totally agree with the suggestion "learn one language and learn it well". But I don't agree with the suggestion "learn the language that is used by your peers". Sure, using the same language comes handy in situations where someone needs some help over a bug. However, it can also be limiting since different languages may provide different solutions to the same problem. So, peers can also benefit from a different mindset that is brought by a different language. Peers should be able to communicate in terms of programming concepts. Let's not forget that algorithms are described to peers in a form of a pseudocode. You gave as an argument the example of Argentina where one should know Spanish in order to communicate. My counterargument would be mathematics. A universal language which many can understand and then translate it into R or Python code. This is my analogy for communicating with programming concepts like "for loops", "if else statements", "functions", "object oriented programming", "functional programming", "hash tables", "data types" etc. So my suggestion would be learn one language, learn it very well and at the same time learn how to express yourself programmatically in a way that can be understood by your peers. And always comment your code.