Today's edition is about deepnote.com, a notebook that will make you love data science.
New here? Subscribe to get the next editions 💌
A word on notebooks
Notebook, notebook, who might you be?
a small book with ruled pages for writing notes in?
Hmm no. We're not going to talk about that kind of notebook. Not today anyway. Instead, we'll discuss data science notebooks, the kind that lives in GitHub repositories. You weren't too far off though, because virtual notebooks were inspired by physical research lab notebooks. 😺
aaand what are virtual notebooks?
Think of them as documents you can use to explore data and share analysis with others. You can usually write Python, R, SQL, or Scala code in them, jot down Markdown-formatted text and even add images! Crucially, notebooks share the ability to run blocks of code independently and display their output.
like that? 👇️
Absolutely! That's fitting when you're trying to answer data questions. It lets you construct hypotheses, test them, analyze the results and showcase the resulting process and conclusions to your audience. Coincidentally, these steps have a lot in common with the scientific method...
Please drop the gobbledygook 🥱 What’s the fuss about notebooks? I mean… Why now?
Because they’re going through an innovation phase that’ll change them forever, and will help data science go mainstream! Notebooks date back to 1988, when Mathematica was introduced, and they need an upgrade. Modern software tools are cloud-based, collaborative by nature, and integrated with other services, so why would notebooks be any different? This very question led to the founding of Deepnote. What follows is their answer.
Collaboration. Reimagined.
If you were designing a modern notebook, what features would you create to help users collaborate? Hint: you’ve seen these features in other products.
Seeing my teammates' avatar when we're on the same notebook?
Granted, Deepnote offers that ✅
You'll even notice them moving across cells, you'll know when they select text and you’ll see them writing code, live!
Viewing a notebook's history and reverting changes made by my colleagues?
I mean, of course that's in there!
Writing comments as I review someone's work?
Yup, that's included too ✅
And there's more! We're talking easy notebook-sharing with public links and granular permissions settings. You name it, it’s in there!
Here’s the good news: I purposely left out plenty of collaborative features. The most curious among you will try Deepnote and feel the excitement of a 5-year old on Christmas Eve when they find them.
Spoiler: entering a notebook and watching someone else run a code cell before your eyes is fascinating. Or maybe that’s just me…
Publishing
Hello beautiful
Creating content is less than half the battle though, whether it's destined for an internal or external audience. If you're looking to create high leverage documents, you need to work on reach and engagement as well. Deepnote thought deeply about this and the result feels like Medium and Jupyter had a baby.
You mean Jupiter right?
No, not the Roman god! That's why it's spelled with a “y”.
Anyway, look at how this chart beautifully interweaves with code and descriptive text. So gorgeous I'm starting to regret my career path.
Sharing is caring
Now you might not be fluent in Python, but I'm sure you understand what it means when a curve skyrockets. In other words, even if you aren't able to write up a notebook just yet, you're able to extract meaning from it. Voilà! You just stumbled on one of the features that make Deepnote so powerful. Not only are Deepnote notebooks charming, but they’re also easily shared with the rest of the world. Yes, including Muggles (an often overlooked problem).
Check out this Face Recognition Demo notebook. I took an existing notebook inspired by Adam Geitgey’s work, duplicated it, and then used Deepnote’s publishing feature to release it to the world. This means you can now open it, read it, and copy it at will. You don’t need to install anything.
Now do this with original content and watch it spread over the web. 🚀
It gets better. All of your published notebooks show up on your profile page, just like for Jakub here. 😉
Your data in a whole new light
Quick start
Forget about mounting data, that’s a relic from the past. With Deepnote you can get started in seconds with an elegant drag & drop interface. Throw in the first csv file you can find and you’re good to go!
Company intel
If you’re getting real about using Deepnote for work, you’ll find integrations aren’t lacking. All it takes is creating a new connection to your company’s data warehouse for Deepnote users to get going (Snowflake and BigQuery are supported out of the box). Once you’re done with that, well… you’re pretty much standing in front of a treasure chest with the biggest chainsaw on the market. What you do with it is up to you. 🤠
Seriously though. I can’t stress enough how powerful this is for a data team. A typical data warehouse contains consolidated data from multiple sources, think Intercom customer support tickets and Stripe orders along with proprietary user behavior data. Combine that with Deepnote and in a matter of minutes you could be investigating how average basket values correlate with refund requests.
Visual visual visual
And if you don’t feel like writing code just yet, Deepnote’s visualize feature is here to help. Simply run a cell with a DataFrame and switch its output type to visualize in order to display it on a chart.
Power when you need it
There’s something in Deepnote for productivity freaks as well, it’s called the Command Palette. It feels like Slack’s own ⌘ + K and opens up just as fast. ⚡️
I’ve lost count of products that offer it. Actually, I’ll start counting those that don’t!
One last thing I had to mention: Deepnote facilitates long-running jobs with machines in the cloud. It won’t interrupt your kernel if you go to sleep while an analysis is running. And if you need greater performance you can always upgrade. Feels like a mini AWS ❤️
Q&A
I'm no data scientist, isn’t learning about this useless to me?
“The more you feel like a noob, the better” said a wise man once. If you’re ready for more, I recommend this course from Dataquest.
How is this better than Google Colab?
Deepnote’s relentless focus on collaboration makes it the go-to choice for teams. You can see other users typing and running code in real-time. Colab does not allow for this kind of concurrent work, nor does it emphasize publishing.
What does the future look like for notebooks?
Hard to say without a palantír. They’re probably going mainstream. They’ll move beyond data teams and reach entire tech organizations first (this has been at work at Neflix for years), and then spread to non-technical users. Deepnote’s next listed features also give us a glimpse as to what’s coming. 👀
Talk soon 👋
Thank you so much for reading! Got questions or want to let me know about products you find exciting? Hit reply.
Talk soon,
Tanguy
Subscribe and get the next editions 💌