I always had the feeling that Java can play a big role in statistics and machine learning. I see no reason against it. And I really don’t like Python. There are multiple reasons, but this post is not about that.
Investigating and exploring data is a type of activity which is non-linear. Ideas pop up from nowhere and you want to be fast enough to follow, to ask questions and receive answers without all the ceremony required when writing a proper piece of production software. Jupyter notebooks comes handy since it allows you to react fast. But Java is not well represented. There are some good kernels, but each lacks something. The one which I liked most, IJava thanks to SpencerPark (https://github.com/SpencerPark/IJava) looks dead for some years with many open issues with no answer.
I will make a confession: I don’t like this type of project. I would favor working on code for data manipulation or clever algorithms. But the interactivity provided by notebook is too good to not pay the price. So I decided to write one. And I decided to pay a share tribute of my free time to all the issues which will be found and write it as best as I can.
You can find it here: https://github.com/padreati/rapaio-jupyter-kernel.
I have to thank to Max Rydahl Andersen for making a better job to collect all those java kernels in a single place and make them available through his amazing https://www.jbang.dev/.
You can find his distribution and java kernels here: https://github.com/jupyter-java/awesome-jupyter-java. I hope this will be the single shop for Jupyter java kernels.
Leave a Reply