So this is going to be a bit outside of the sphere of data science, but I think it's worth sharing anyway. I listened this podcast a few weeks ago.

The podcast got me pretty excited because nbdev basically introduces a new way of writing software. Instead of using an IDE and placing your code in .py files, it changes the whole approach and lets you develop even standard software packages inside notebooks.

So I decided to try it out in Deepnote. So here's a published notebook with a guide and an example of how nbdev works: https://deepnote.com/@the21st/nbdev-Deepnote-MQQVcO8aQV--6rCKUm-ftg

It's a very basic example, but it works! Let me know what you think 🙂

As you might know, we've recently added a new way to visualize DataFrames to Deepnote: chart blocks (cells).

When I was working in this feature, I looked at other similar tools in the ecosystem so I decided to write a short article comparing the most relevant ones. Check it out!


Alternatives to matplotlib for pandas dataframes

What are the best tools for no-code pandas visualization tools and how do they compare?

https://medium.com/deepnote/alternatives-to-matplotlib-for-pandas-dataframes-f15be61c3cf4


Let me know if there are other ones out there I might have missed 🙂

It's not much, but it's starting to look solid 😎

https://deepnote.com/@the21st

Hey there! For those of you who missed it in the product update, I created a video showcasing a new type of cell/block in Deepnote – chart cells:



Chart blocks are really great for rapid prototyping of charts because you don't need to type any code! I love tools like Tableau or Metabase and I missed their UI charting capabilities in notebooks so I was happy I could work on this new feature ☺️

This is me: https://www.linkedin.com/in/simon-sotak/

I haven't created a CV in a while, I thought everyone just uses LinkedIn.

One thing I'm not happy with is that Represent was acquired by Custom Ink so I didn't change my job in 2017, but I'm not sure how to explain that on LinkedIn.

Roast my linkedin!

I saw this cool project named SkinDeep by Vijish Madhavan trending on reddit today. So I decided to try and port it to Deepnote, and it was super easy, runs on our free CPU machines pretty fast, check out the result:



Here's the project if you want to duplicate it and give it a shot:

https://deepnote.com/project/SkinDeep-D8BAqIJ1RLKh2M0sC6s4cg/%2FSkinDeep.ipynb

I stumbled upon this cool website that visualizes real-world loss functions and lets you simulate gradient descent with adjustable learning rates. Pretty rad. I always visualized this only in my head 😄

https://losslandscape.com/explorer

Hi everyone, 

Deepnote is currently experiencing issues with loading and running projects. We are investigating and will hopefully have a fix soon.

Hang in there!
Simon