I try to get an overview about the most recent discussions of how machine learning and classical statistics differ.

There is an excellent discussion on this issue here on stackexchange ( The Two Cultures: statistics vs. machine learning? ), but the original posts are about 8 years old by now.

Does anyone know of a more recent discussion (here on this forum or elsewhere)? Or are the main points in the Breiman paper still valid?


I don't go into technical details, but there are some relevant articles on my blog fharrell.com, and today I'll be posting a new article giving overall guidance in making the choice between statistical models and machine learning. I'd welcome discussion about that article, either on the blog or here.

  • $\begingroup$ In your post, can you link directly to some articles you feel are good examples? $\endgroup$ – Cam.Davidson.Pilon Apr 30 '18 at 12:21
  • $\begingroup$ I have not done a good enough job surveying the literature to be able to do that at present, but I will include 2 interesting references. We need more. $\endgroup$ – Frank Harrell Apr 30 '18 at 12:59
  • $\begingroup$ Ah, I was unclear in my comment. By "post", I was referring to this post above, and by "articles", I meant specifically your blog articles. $\endgroup$ – Cam.Davidson.Pilon Apr 30 '18 at 22:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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