Hi everyone! I've released kstats, a new Kotlin Mu...

# datascience

[Pavel Gorgulov]

Hi everyone!
I've released kstats, a new Kotlin Multiplatform statistics library.

It includes descriptive statistics, probability distributions, hypothesis tests, correlation analysis, regression, and sampling utilities.

For more details:
• Release notes: https://github.com/Oremif/kstats/releases/tag/0.3.0
• Documentation: https://kstats.oremif.org
Slack Conversation

[Michal Harakal]

Looking forward to see how you implement deeper integration with Kotlin-Notebook 🙂

[Pavel Gorgulov]

I’ve already added kstats to the notebook descriptors:
https://github.com/Kotlin/kotlin-jupyter-libraries/blob/master/kstats.json

it can now be used in notebook like this:

%useLatestDescriptors()
%use kstats

[Michal Harakal]

This is a nice start. I’ll try that soon.
I was also thinking about bringing a kotlin-notebook wrapper using the Kotlin Jupyter API for my library, distributed via Maven Central as a single JAR dependency.
The main reason is to avoid relying on descriptors from GitHub. I remember having issues during a DataFrame training session when multiple participants were hitting GitHub’s quota limit. That’s not too critical for me right now, but it is still something I’d prefer to avoid.
I’m also planning to add some callbacks on startup. I already tried this once and failed, so I plan to revisit it later.
For now, going with a JSON descriptor seems like a low-hanging fruit to improve the user experience.

[zaleslaw]

Agree, descriptors on the Github could be a bottleneck