6

Generalized Additive Models [Trevor Hastie and Robert Tibshirani 86] was well received with over 1335 Citations.

I am also aware of the popular(?) version of GAM - the Multivariate Adaptive Regression Splines MARS by Friedman 91 - 4568 Citations.

I would like to know how popular / scalable / practically valid, the sparse version of GAM - the Sparse Additive Model ( SpAM ) is in comparison to the MARS. I am planning to implement this for academic purposes. Would it be worth the effort spent in learning about this assuming I am trying to learn more about GAMs in general or should I choose any other state of the art GAM model which is widely accepted ( Other than MARS ) to implement and learn?

It would be awesome if you could keep in mind that I am a newbie to this field.

Also I am new to this community please let me know if this question is considered on topic too.

References for SpAM:

Also closely related - Generalized additive models -- who does research on them besides Simon Wood?

EDIT:

To summarize these are the 3 questions I'd love an answer for -

  • With just academic interest and no particular application in mind, what is the state of the art research in GAMs?
  • How popular / scalable / practically valid is the sparse version SpAM?
  • I'd like to learn about GAMs and implement a GAM based model that is widely accepted ( like MARS ) preferably in Python ( MARS already has PyEarth ). Which one would you recommend and why?
Community
  • 1
Raghav RV
  • 111
  • 8
  • 1
    The paper you are presumably alluding to is by Trevor Hastie and Robert Tibshirani in _Statistical Science_. You omitted the first author and misspelled the name of the second. (Anyone answering this question would know that, but it's put in for some others.) – Nick Cox Dec 01 '15 at 16:01
  • 2
    I suggest that you are posing three very different questions. The first alone calls for a review paper. Are new implementations needed? (Whether MARS is a "popular version" of GAM I leave to experts.) You are posting as a newbie aiming to offer an implementation for researchers; that is impressively ambitious. – Nick Cox Dec 01 '15 at 16:05
  • "The ... others." - Thanks! Have edited the same :) ; "You are posting as a newbie aiming to offer an implementation for researchers;" - Ah! nope ;) Like I said I'd like to "to implement and learn". The reason I was looking for a good one to implement is so that I don't expend my energy in a wrong direction :) – Raghav RV Dec 01 '15 at 17:01
  • 2
    I like your thoughtful questions but it will be closed as too broad. Divide the three questions into separate questions and it will help to make them more precise. – Michael R. Chernick Aug 22 '18 at 16:28

0 Answers0