dear all, I want to choose between different polynomial robust regressions, however R2 may be misleading since the highest de degree is, the "better fit" to data. Is there any function implemented in robust or robustbase that computes an index for "model selection" ??? Thanx for helping! pep -------------- next part -------------- A non-text attachment was scrubbed... Name: josep_serra.vcf Type: text/x-vcard Size: 448 bytes Desc: not available URL: <https://stat.ethz.ch/pipermail/r-sig-robust/attachments/20101201/e6b5a985/attachment.vcf>
[RsR] best robust fit
3 messages · Samuel Le, Pep Serra
Have you thought of the Akaike Information Criterion? It is the R squared penalized by the number of regressors. You can access to it using reg$AIC if reg is your regression object. Samuel -----Original Message----- From: r-sig-robust-bounces at r-project.org [mailto:r-sig-robust-bounces at r-project.org] On Behalf Of Pep Serra Sent: 01 December 2010 14:48 To: r-sig-robust at r-project.org Subject: [RsR] best robust fit dear all, I want to choose between different polynomial robust regressions, however R2 may be misleading since the highest de degree is, the "better fit" to data. Is there any function implemented in robust or robustbase that computes an index for "model selection" ??? Thanx for helping! pep __________ Information from ESET NOD32 Antivirus, version of virus signature database 5663 (20101201) __________ The message was checked by ESET NOD32 Antivirus. http://www.eset.com __________ Information from ESET NOD32 Antivirus, version of virus signature database 5663 (20101201) __________ The message was checked by ESET NOD32 Antivirus. http://www.eset.com
I am not an expert, but I think AIC is not suitable when using robust statistics, maybe someone smarter may give a hint on this I checked http://www.mail-archive.com/r-help at stat.math.ethz.ch/msg39949.html Thank you anyways and please do not hesitate to respond if you have further comments... pep Al 01/12/2010 15:51, En/na Samuel Le ha escrit:
Have you thought of the Akaike Information Criterion? It is the R squared penalized by the number of regressors. You can access to it using reg$AIC if reg is your regression object. Samuel -----Original Message----- From: r-sig-robust-bounces at r-project.org [mailto:r-sig-robust-bounces at r-project.org] On Behalf Of Pep Serra Sent: 01 December 2010 14:48 To: r-sig-robust at r-project.org Subject: [RsR] best robust fit dear all, I want to choose between different polynomial robust regressions, however R2 may be misleading since the highest de degree is, the "better fit" to data. Is there any function implemented in robust or robustbase that computes an index for "model selection" ??? Thanx for helping! pep __________ Information from ESET NOD32 Antivirus, version of virus signature database 5663 (20101201) __________ The message was checked by ESET NOD32 Antivirus. http://www.eset.com __________ Information from ESET NOD32 Antivirus, version of virus signature database 5663 (20101201) __________ The message was checked by ESET NOD32 Antivirus. http://www.eset.com
-------------- next part -------------- A non-text attachment was scrubbed... Name: josep_serra.vcf Type: text/x-vcard Size: 448 bytes Desc: not available URL: <https://stat.ethz.ch/pipermail/r-sig-robust/attachments/20101201/545753e2/attachment.vcf>