The video is recorded on Youtube:

You can see the slides here: slides

The talk itself is based on this notebook that I published on this blog yesterday and that I used to demo during the talk.

The host of the conference was Istvan Hajnal. He tweeted the following:

He also took the R out of my family name NachteRgaele. Troubles with R, it's becoming a story of my life... :joy: Behind the scene Kris Peeters calmly took the heat of doing the live streaming. :+1: Almost Pydata quality! Big thanks to the whole Data Science Leuven team that is doing all this on voluntary basis.

Standing on the shoulders of the giants

This talks was not possible without the awesome Altair visualisation library made by Jake VanderPlas. Secondly, it builds upon the open source Shap library made by Scott Lundberg. Those two libraries had a major impact on my daily work as datascientist at Colruyt group. They inspired me in trying to give back to the open source community with this talk. :metal:

If you want to learn how to use Altair I recommend the tutorial made by Vincent Warmerdam on his calm code site: https://calmcode.io/altair/introduction.htm

I would also like to thank my collegues at work who endured the dry-run of this talk and who made the suggestion to try to use a classifier to explain the clustering result. Top team!

Awesome fastpages

Finally, this blog is build with the awesome fastpages. I can now share a rendered Jupyter notebook, with working interactive demos, that can be opened in My binder or Google Colab with one click on a button. This means that readers can directly tinker around with the code and methods discussed in the talk. All you need is a browser and an internet connection. So thank you Jeremy Howard, Hamel Husain, and the fastdotai team for pulling this off. Thank you Hamel Husain for your Github Actions. I will cast for two how awesome this all is.