I am currently working in BigData with Spark-Scala framework. I want to learn Machine learning from scratch. Which language would be better to learn for machine learning, Scala or Python?

  • $\begingroup$ Check this datascience.stackexchange.com/questions/10884/… . This could be thought as an opinion based question, but I would recommend python too for the reasons that Feliz describes. One more point would be that the transition is easy from whatever language you have background on. $\endgroup$ – Nikos Jul 18 '19 at 7:31

I have done only a little bit of scala (spark streaming for database replication), but have worked with python for about a year. I would recommend python for the following reasons:

  • We do a lot of notebook centric development at my job. I would recommend that because the Jupyter IDE allows you to visualize/debug your data really effectively, which is important to understanding and validating concepts.

  • Python is dynamically typed and has a powerful set of built-ins and utilities that will let you transform your data into the shape you need pretty easily.

  • Python has first-class support for almost all ML libraries (TensorFlow, PyTorch, Sci-kit Learn).

| improve this answer | |
  • $\begingroup$ Thanx Feliz..That explanation gives much clarity.!! :) $\endgroup$ – Siraj Jul 18 '19 at 9:41
  • $\begingroup$ Can you also tell me some source which will be good to learn Python intended for Machine Learning.? $\endgroup$ – Siraj Jul 18 '19 at 9:47
  • $\begingroup$ To be honest, I ended up having to pick it up on the job. One talk I found to be exceptional was by David Beazley about "builtin superheroes". Very relevant in my day to day when it comes to cleaning, aggregating, and grouping of data for analysis: youtube.com/watch?v=lyDLAutA88s $\endgroup$ – FelizNaveedad Jul 22 '19 at 0:42

Not the answer you're looking for? Browse other questions tagged or ask your own question.