Monday, September 28, 2015

I need some sympathy: spending most of my time coding in Java and Python

As the universe unfolds, I have been spending most of my time recently working with machine learning and for the forseeable future that will not change. Lets face it, many of the really great ML libraries are written in Java and Python.

I still love development with Clojure and Ruby, and I am still on my long term quest to become passably proficient with Haskell. That said, it is crazy to not simply use languages that have the best library and framework support for any task.


Dale Asberry said...

There's always Spark with MLlib and Scala!

Mark Watson, author and consultant said...

Hello Dale,

I have been using Spark and MLlib a lot lately, mostly because I use MLlib heavily in a new book I am wriitng. For the book ( MLlib was a great fit because readers have a good dev mode running Spark locally, with an easy path to scaling as needed.

gozek said...

Isn't using Java APIs from Clojure convenient? Is better to use Java directly?

Mark Watson, author and consultant said...

Hello Gozek,

It is easy to call Java from Clojure, buch easier than calling Java from Scala (I think).

Still, if you are using large libraries and adding just a bit of code for data import, cleaning, etc. it is probably just quicker user Java.

Federico Tomassetti said...

Perhaps you can give a try to Frege (a.k.a. Haskell for the JVM)