Dear all, I want to do a multivariate regression. So I have Y which is a matrix and one vector x which is my predictor variable. I want to do a multivariate regression with penalizing the coefficients I get. Something like: $||y-xb|| + \lambda b^t P b $ But I have a "own" penalty term which I want to use for penalized regression. When I searched for penalized regression, I found a lot of packages but all of them have predefined penalties like lasso, ridge or second differences. I used the gam() package in mgcv when I had an univariate response: gam(Y~x_1, paraPen = penaltymatrix) but this package do not support multivariate regression. Is there any R package where I can use an individual penalty matrix to my coefficients? Or can you give me any advice how I solve the problem I have? Kind regards, Dominik [[alternative HTML version deleted]]
"but this package do not support multivariate regression." Wrong. "Fits a generalized additive model (GAM) to data, the term ?GAM? being taken to include any quadratically penalized GLM and a variety of other models estimated by a quadratically penalised likelihood type approach (see family.mgcv <http://127.0.0.1:24757/help/library/mgcv/help/family.mgcv>). " So quadratic penalties only. Also: This list is about R programming. Questions about statistical methodology are generally off topic. Try stats.stackexchange.com instead for that. Bert Gunter "The trouble with having an open mind is that people keep coming along and sticking things into it." -- Opus (aka Berkeley Breathed in his "Bloom County" comic strip ) On Mon, Apr 29, 2019 at 5:07 AM Dominik Schmidt <schmidtdominik22 at gmail.com> wrote:> Dear all, > > I want to do a multivariate regression. So I have Y which is a matrix and > one vector x which is my predictor variable. I want to do a multivariate > regression with penalizing the coefficients I get. Something like: > $||y-xb|| > + \lambda b^t P b $ But I have a "own" penalty term which I want to use for > penalized regression. > > When I searched for penalized regression, I found a lot of packages but all > of them have predefined penalties like lasso, ridge or second differences. > > I used the gam() package in mgcv when I had an univariate response: > > gam(Y~x_1, paraPen = penaltymatrix) > > but this package do not support multivariate regression. > > Is there any R package where I can use an individual penalty matrix to my > coefficients? Or can you give me any advice how I solve the problem I have? > > Kind regards, > > Dominik > > [[alternative HTML version deleted]] > > ______________________________________________ > R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide > http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >[[alternative HTML version deleted]]
I should have added: for multivariate gam models see e.g. ?mvn Bert Gunter On Mon, Apr 29, 2019 at 7:51 AM Bert Gunter <bgunter.4567 at gmail.com> wrote:> "but this package do not support multivariate regression." > Wrong. > > "Fits a generalized additive model (GAM) to data, the term ?GAM? being > taken to include any quadratically penalized GLM and a variety of other > models estimated by a quadratically penalised likelihood type approach (see > family.mgcv <http://127.0.0.1:24757/help/library/mgcv/help/family.mgcv>). > " > > So quadratic penalties only. > > Also: This list is about R programming. Questions about statistical > methodology are generally off topic. Try stats.stackexchange.com instead > for that. > > Bert Gunter > > "The trouble with having an open mind is that people keep coming along and > sticking things into it." > -- Opus (aka Berkeley Breathed in his "Bloom County" comic strip ) > > > On Mon, Apr 29, 2019 at 5:07 AM Dominik Schmidt < > schmidtdominik22 at gmail.com> wrote: > >> Dear all, >> >> I want to do a multivariate regression. So I have Y which is a matrix and >> one vector x which is my predictor variable. I want to do a multivariate >> regression with penalizing the coefficients I get. Something like: >> $||y-xb|| >> + \lambda b^t P b $ But I have a "own" penalty term which I want to use >> for >> penalized regression. >> >> When I searched for penalized regression, I found a lot of packages but >> all >> of them have predefined penalties like lasso, ridge or second differences. >> >> I used the gam() package in mgcv when I had an univariate response: >> >> gam(Y~x_1, paraPen = penaltymatrix) >> >> but this package do not support multivariate regression. >> >> Is there any R package where I can use an individual penalty matrix to my >> coefficients? Or can you give me any advice how I solve the problem I >> have? >> >> Kind regards, >> >> Dominik >> >> [[alternative HTML version deleted]] >> >> ______________________________________________ >> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see >> https://stat.ethz.ch/mailman/listinfo/r-help >> PLEASE do read the posting guide >> http://www.R-project.org/posting-guide.html >> and provide commented, minimal, self-contained, reproducible code. >> >[[alternative HTML version deleted]]