Thread by @mdancho84

BREAKING NEWS: #ChatGPT Breaks #Python VS #R Barriers For #DataScience Teams Everywhere...

And Data Scientists everywhere are in shock.

Here's the LIVE story as it's unfolding.🧵

#rstats #pydata

It's NOT #R VS #Python ANYMORE!

Let me explain with an example that was MIND BLOWING to me.

Here's the background:

I'm an R guy.

I use R for research.

It's quick to make analysis.

And now I'm super good with it.

But I work with 9 Python data scientists.

And it's a pain in the butt because Python is not my first choice.

And my python coworkers hate R.

So there's a natural divide.

But, what if there was a way to work together more easily.

Here's how ChatGPT just solved this problem.

We have a Customer Sales Prediction project going on...

And my weapons of choice for experimental business research are:

- #R (stats-forward programming language)
- #tidyverse (data analysis)
- #Tidymodels (machine learning)
- #XGBoost (top-ranked algorithm)

So I asked chatgpt to create a model for me using tidymodels and xgboost to predict customer spending value...

Feck me. 🤯

#Chatgpt just made the ML model in #R with #tidymodels.

But again, I could do this in about 30 minutes.

But what I really want help with is delivering something to my Python team...

So they can use my model in production!

What is "Production"?

"Production" is just a fancy word for giving others in the company access to it so they can make predictions on new customers on-demand.

So it's super important to add business value.

In the past it would take me a full day to do this translation.

Full Disclosure:

After pounding my head at Python for years, I am much better now.

But learning Python was not fun.

And what I'm about to show you is a game changer...

(and much more fun)

So next, to get my model into Python for production code, I asked chatgpt to convert my R script to Python with Pandas and Scikit Learn.

(cuz that's what my team uses).

Feck me twice.

Chatgpt just saved me 2 hours of work.

But this #python code is pretty much useless...

Unless I can put it into an API.

What's an API?

Glad you asked.

An API is just a way for business people to connect to my model.

They can use tools like Excel, Tableau, PowerBI, to connect to my model and get predictions on demand.

So I asked ChatGPT, "Can you make me an API with FastAPI?"

Because that's what my team likes.

WTFeck! It just made the FastAPI for me!

(my team is going to love me)

But I have one last problem.

I know my team needs to have a #Docker image so they can put that API into production.

So I decided to help them out.

I asked ChatGPT...

"Can you create a docker file to expose the API?"

And ChatGPT delivers the DockerFile! 🤯

So to recap:

1. I'm an R guy

2. I work with a bunch of Python guys and gals

3. They want Python code (ChatGPT just gave me the python code they need)

4. They want production (ChatGPT just gave me the API + the DockerFile they need)

5. And I did this in 10 minutes.

If you're reading this, you are either in 1 of 2 states right now:

1. Super excited (mind blown)

2. Super freaked out (my job is gone)

Well, let me help you turn both of these feelings into something productive.

Over the past 4 weeks, I've been quietly researching ChatGPT for data science.

And I'm now ready to reveal the results.

What's the next step?

Register for my free ChatGPT for Data Science training

And gain 5 more recipes for applying chatgpt to data science.

Register Here: us02web.zoom.us/webinar/regist…