1
$\begingroup$

I would like to fit a nonparametric regression model with two predictors. What R packages are available? I found 'loess' function in 'stats' package and 'gam' function in 'gam' package. Is there anything else?

Improve this question
$\endgroup$
3
  • 2
    $\begingroup$ Depending on what you mean by non-parametric, boosted regression using the gbm package might help. $\endgroup$
    – Jeremy Miles
    Commented Aug 12, 2014 at 17:58
  • $\begingroup$ Thanks Jeremy. What I would like to do is basically to compare multiple methods; so I need as many methods as possible. Please recommend anything if you know something. $\endgroup$
    – user67275
    Commented Aug 12, 2014 at 18:26
  • 1
    $\begingroup$ If you want something a bit "exotic" you might try multivariate adaptive regression splines as those are available in the package earth. $\endgroup$
    – usεr11852
    Commented Aug 12, 2014 at 21:05

1 Answer 1

Reset to default
2
$\begingroup$

This is far from comprehensive, but you should also take a look at locfit for various "local" methods and wavethresh for wavelet methods. John Fox's list of packages from 2005 should be helpful (though is bound to be somewhat outdated).

Improve this answer
$\endgroup$
2
  • $\begingroup$ Thank you for your suggestion. I just tried 'locfit' and it works well. But I don't know which function to use in 'wavethresh'. Could you give me a very simple example? (In case of 'locfit', it would be locfit(Y~X1+X2,data=data.sample) ) $\endgroup$
    – user67275
    Commented Aug 14, 2014 at 18:07
  • $\begingroup$ I don't know wavelet methods so well so I'll just point you to Schafer and Wasserman's tutorial on nonparametric inference (example using 'wavethresh' starts on pg. 92). You can get the WMAP data from the example at Wasserman's website. Hope that helps! $\endgroup$
    – daniel.s
    Commented Aug 15, 2014 at 7:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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