Tip #58: Automated data analysis in Colab?


Hi Reader,

Last week, I invited you to help me test Google's Data Science Agent in Colab, which promises to automate your data analysis.

Does it live up to that promise? Let's find out! 👇


Sponsored by: Morning Brew

Business news you’ll actually enjoy

Join 4M+ professionals who start their day with Morning Brew—a free daily newsletter that makes business, tech, and finance news genuinely enjoyable to read and hard to forget. Each morning, it breaks down complex stories in plain English—cutting through the noise with sharp insights and just enough wit to keep you engaged. In under 5 minutes, you’ll be up to speed on what matters most to your career and the world around you—all before you’ve finished your morning coffee.


🔢 How to use Data Science Agent in Colab:

Before we get to the results, here's a recap of how it works:

  1. You open a blank Colab notebook.
  2. You upload a data file.
  3. You describe what analysis you want done.
  4. Gemini does the analysis for you.

The interface is a bit confusing, so I recorded a short video (no audio) to help you get started.


🎯 Does it do a good job with the analysis?

Here's what tester #1 said:

I tested it with a non-trivial statistical analysis and I should say... the results are really impressive. Implementing the same code from scratch, without an existing pipeline, it would have taken to me more than one hour (to be optimistic!!)

Tester #2:

This Data Science Agent in Colab is so powerful! I created a mock dataset for testing and asked it to calculate the cluster coherence. Then it came out with a plan and executed it. Most amazing part is that it installed the missing packages by itself when running into an error. Finally, it did answer my question (Which cluster has the smallest coherence value?) which is amazing.

Tester #3:

It was good at simple analysis but might not work great when given complex problems.

👉 Here are my takeaways:

After reviewing the Colab notebooks shared by testers and testing it myself, here's my overall conclusion:

Data Science Agent is only useful if you are able to evaluate whether the steps it takes are correct. It will come up with a plan and write the code to execute that plan, but you still need to know enough to assess:

  • Is it including all of the steps that are necessary to solve this problem?
  • Is it making reasonable assumptions?
  • Is it ignoring any relevant factors?
  • Is the code it writes in alignment with the stated goal?

As such, Data Science Agent is most useful for those who could already complete the analysis on their own, but just want help in order to execute the analysis faster.

Thus if you use Data Science Agent without sufficient expertise, you run the risk of performing a misleading (or incorrect) analysis!


See you Friday! 👋

If you enjoyed this week's tip, please considering sharing it with a friend!

I'll be back in your inbox on Friday to share the top AI news of the week.

- Kevin

P.S. My favorite AI-generated video so far 😂

Learn Artificial Intelligence from Data School 🤖

Join 25,000+ intelligent readers and receive AI tips every Tuesday!

Read more from Learn Artificial Intelligence from Data School 🤖

Hi Reader, In this week’s tip, I’ll be breaking down some highly practical advice for taking full advantage of the capabilities of today’s AI models. Check it out below! 👇 Today’s tip is based on Ethan Mollick’s excellent article, Using AI Right Now: A Quick Guide. I recommend reading the whole thing if you have time, but if not, I’ve pulled out some important quotes from the article and added my own commentary: For most people who want to use AI seriously, you should pick one of three...

Hi Reader, Here are the most important AI stories I’ve found this month: Microsoft’s AI solves medical mysteries Cloudflare charges AI crawlers Books are legal for AI training AI model teaches itself AI researchers are paid like superstars Details below! 👇 🩺 Microsoft’s AI solves medical mysteries Microsoft’s new “AI Diagnostic Orchestrator” solved 85% of 304 real medical mysteries, whereas experienced doctors (who did not have access to colleagues or medical databases) only solved an average...

Hi Reader, Here are your top AI stories for the week: ChatGPT can weaken your brain Claude shares nerve gas recipe Amsterdam ends AI experiment due to bias Read more below! 👇 Sponsored by: Brain.fm Transform Your Focus With Brain.fm I know you're always on the hunt for tools that genuinely improve your life—which is why I'm excited to introduce you to Brain.fm's groundbreaking focus music. Brain.fm's patented audio technology was recently validated in a top neuroscience journal, showing how...