Related Posts
Bowl Leader
Welcome media friends!
Do I need to know ratios for FDD case study?
Conversation Starter
Additional Posts in Data & Analytics Consultants
Community Builder
Got messaged by a C3 . ai recruiter. Read that wlb is bad and that the interview process is absurdly long, but the Glassdoor reviews are 4.2 and can't find actual hours worked posted by anyone. How's the culture really? I'd be aiming for DS consulting, something more functional but with DS/ML concepts as my differentiator.
C3.ai, Inc.
New to Fishbowl?
Download the Fishbowl app to
unlock all discussions on Fishbowl.
unlock all discussions on Fishbowl.
If you are able to read code and make sense out of it, go to chatgpt convert your python code to R code, run and check if gets the same results, if not, try changing it until you get the same result
SC1: fair enough, we’ve just had a lot of hoopla about LLM’s going around the firm recently
I personally like R better but it depends on the overall skill set of the team and what you are trying to accomplish. The Tidyverse libraries make things very simple for data manipulation. Rstudio is a great IDE. R shiny is more mature than Dash for dashboard.
Python is better for any deep learning applications and NLP generally. Some people with a programming background thing Python is more intuitive but personally I think it’s OOP features add a lot of extra steps to pipelines. I think Python package management is a huge pain as well.
What Numpy/Pandas does after importing them to Python, R does that natively. I always like R and default to it if I have the choice. And R Studio is a fantastic IDE. R is the statistician’s choice.
Python just has access to so many packages and the documentation is great. Deep learning is better in Python.
Use R if you never want to work on interesting things ever again.
Hey I’m looking for ML/AI/software engineers that code in R… said no one, ever.
I like R more because RStudio is so good. But it doesn't matter anymore, GPT4 will give you the code you need.
I wrote 500-600 lines of code in about 1.5 hours using ChatGPT for a side project.
I think it's all about knowing the stats and evaluating models, several tools will write the code for you.
So many tools can write code for you, it’s about knowing the process and what code goes where now.
I prefer python for development, but that is because we are using DL for our clients. Python skills are more favourable in my practice than R, but I assume if you picked one up you can figure out the other
Imagine using r in prod
I used R in prod at a big tech. It's got all the OO and devops stuff you need, contrary to what reading scripts by academics would have you believe
Python has a more consolidated package ecosystem, much better community based documentation, fringe ML implementations, and many more job postings.
R has slightly more convenient data exploration, much easier viz, fringe quant implementations, and often more detailed documentation.
Both have standard stuff you need to do 99% of your work, engineering tooling, and plug into C++ for speedups if you reaaaally wanna go there.
Develop POCs with Python and hire people who know C++ for implementation
Visual Storyteller
Follow Matt Dancho on LI for this exact same argument and comparison - he leans R
I use both as they both have a lot of similar strengths, but also some differences. If you’re starting out these differences don’t matter so much.
Start with R if you plan to do a lot of stats stuff. If you don’t know what you will need the language for, then you can start with Python. Then learn the other when you need to (eg can’t do sth easily in one language check to see if you could in the other). Ince you know one language well it’s relatively easy to hop across languages.
Python gives you more flexibility to do more with your data. R might have some nice libraries and functions but python has all of that and more.