2
$\begingroup$

There are dozens of algorithms one can use to build a predictive model.

What books or studies exist that can help one determine which algorithm to use? Elements of Statistical Learning spends a lot of time talking about each model it covers (and it covers a lot of them), but not much time comparing use cases and performance.

$\endgroup$
  • 2
    $\begingroup$ Conventionally, one tests several alternatives and compare out-of-sample performance metrics. $\endgroup$ – Sycorax says Reinstate Monica Jul 16 '15 at 13:39
2
$\begingroup$

Elements of Statistical Learning is a great book. If you read and understand the material, you should be able to discern what algorithms might be better suited for what type of problem. If you'd like a simpler version of that book, try An Introduction to Statistical Learning from the same authors. This simpler book focuses more on practical application, providing a comparison of the different methods in the book. It's less comprehensive than the ESL book though.

This diagram from SciKit-Learn is a pretty good one for illustrating machine learning possibilities.

But as User777 has already pointed out, usually we narrow down to a few possible procedures and use some sort of cross validation method to test the out-of-sample error rate.

$\endgroup$
  • $\begingroup$ (+1) That flowchart is going to come in handy. $\endgroup$ – ebb-earl-co Jul 16 '15 at 14:15

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.