<https://www.crowdcast.io/e/data-science-on-the-jv...

# datascience

[Pratik Patel]

https://www.crowdcast.io/e/data-science-on-the-jvm

[altavir]

Sadly, it requires singn-in to watch

[Pratik Patel]

hm

[altavir]

Ah, OK google account is working

[Pratik Patel]

cool. it's the first time I have given this presentation, so please let me know if there's any feedback. I plan to expand this into a 3 to 5 part series with hands-on workshop.

[altavir]

I do not think I will be using Zeppelin anytime soon, But I am really looking forward to a plugin framework for Kotlin kernel.

[Pratik Patel]

I agree that Zeppelin is suitable for a narrow number of real-world use-cases. But I think that it is a easy and interactive way to get folks interested in the topic.

[altavir]

Yes, indeed

[Pratik Patel]

If you have some github repo's that have more examples, please send them my way. I found very little complete examples to compare what I was doing with others.

[altavir]

I am maintaining plotly.kt library. which have experimental Jupyter-lab integration. Sadly, the main problem is the loading of companionon JS bundle. So, I had to do some hacks and it probably won't work on the other platforms. Now I am finalizing much mode advsnced VisionForge library, which will incorporatr plotly integration and will be able to do much more. Still, JS bundle integration is the major problem.

[Michal Harakal]

@Pratik Patel Hi, I liked your presentation. Were you able to solve the issues with response times? It was really slow. I don't think we can convince python DS people with this. I am sure that its tehcnical issua of early versions and this wont stop me. Maybe its just dependecy resolver running for the first time etc....

[Pratik Patel]

@Michal Harakal It was because I was livestreaming from the browser, and my macbook's battery/charger was not providing enough power so the CPU was being throttled. Normally Zeppelin is very fast to run code/paragraphs

[Michal Harakal]

Cool, thank you, good to know....