search for: lpmatrix

This is in regards to the suggested use of type="lpmatrix" in the documentation for mgcv::predict.gam. Could one not get the same result more simply by using type="terms" and interpolating each term directly? What is the advantage of the lpmatrix approach for prediction outside R? Thanks. -- View this message in context: http://www.nabble....

The documentation for predict.gam in library mgcv gives an example of using an "lpmatrix" to do approximate prediction via interpolation. However, the code is specific to the example wrt the number of smooth terms, df's for each,etc. (which is entirely appropriate for an example) Has anyone generalized this to directly generate code from a gam object (eg SAS or C code)? I wa...

...ine fit, I think I would just need the coefficients and the basis generating function. I've been looking at Chapter 5 (p243-247) in Wood (2006) "GAMs: An Introduction with R", and I've been trying to understand what is needed for the smooth.construct and the predict.gam(type='lpmatrix') functions. It's possible to generate a new prediction matrix with predict(gm, newdata, type='lpmatrix), but the entire model object is needed as input. I looked into it further, and the predict method (using type='lpmatrix") depends on the following model components: ob...

...] S<-matrix(0,19,19) for (i in 2:10) { for (j in 2:10) { S[i,j]=S1[i-1,j-1] S[i+9,j+9]=S2[i-1,j-1] } } beta<-m1$coef #penalised deviance Dp<-m1$dev+t(beta)%*%S%*%beta F1<-Dp/(2*phi) #2nd term F2<-sum(ifelse(y==0,0,y*log(y)-y-log(factorial(y)))) #3rd term X<-predict(m1,type="lpmatrix") W<-diag(fitted(m1)) H<-t(X)%*%W%*%X ldhs<-determinant(H+S,log=TRUE)$modulus[1] eigS<-eigen(S,only.values=TRUE)$val lds<-sum(log(eigS[1:16])) F3<-(ldhs-lds)/2 #4th term Mp=3 F4<-Mp/2*log(2*pi*phi) F1-F2+F3-F4 m1$gcv #reml score = formula ##quasipoisson with scale = 1 #...

Dear R-Users, Dr. Wood replied to a similar topic before where confidence intervals were for a ratio of two treatments ( https://stat.ethz.ch/pipermail/r-help/2011-June/282190.html). But my question is more complicated than that one. In my case, log(E(y)) = s(x) where y is a smooth function of x. What I want is the confidence interval of a ratio of log[(E(y2))/E(y1)] given two fixed x values of

...am function under mgcv and I thought that with the gam function I could obtain H using s(). Is this wrong? e.g. mod_spline<-gam(yy~s(x.summer,k=16,bs="cr",fx=FALSE,by=NA,m=1),sp=c(0.001),data=dstime,family=poisson,fit.method="mgcv") H<-predict.gam(mod_spline,type="lpmatrix") H<-H[,2:16] I presume that the vectors obtained using the predict.gam() are different from the ones obtained using GAM.setup(). Thank for your help. Looking forward to hearing from you. Michela [[alternative HTML version deleted]]

...,j]=S1[i-1,j-1] > S[i+9,j+9]=S2[i-1,j-1] > } > } > beta<-m1$coef > #penalised deviance > Dp<-m1$dev+t(beta)%*%S%*%beta > F1<-Dp/(2*phi) > > #2nd term > F2<-sum(ifelse(y==0,0,y*log(y)-y-log(factorial(y)))) > > #3rd term > X<-predict(m1,type="lpmatrix") > W<-diag(fitted(m1)) > H<-t(X)%*%W%*%X > ldhs<-determinant(H+S,log=TRUE)$modulus[1] > eigS<-eigen(S,only.values=TRUE)$val > lds<-sum(log(eigS[1:16])) > F3<-(ldhs-lds)/2 > > #4th term > Mp=3 > F4<-Mp/2*log(2*pi*phi) > > F1-F2+F3-F4 &gt...

Hi folks, I originally tried R-SIG-Mixed-Models for this one (https://stat.ethz.ch/pipermail/r-sig-mixed-models/2010q3/004170.html), but I think that the final steps to a solution aren't mixed-model specific, so I thought I'd ask my final questions here. I used gamm4 to fit a generalized additive mixed model to data from a AxBxC design, where A is a random effect (human participants in

I have a question about the basis functions of cubic regression spline in mgcv. Are there some ways I can get the exact forms of the basis functions and the penalty matrix that are used in mgcv? Thanks in advance! Yan [[alternative HTML version deleted]]

...ements of the variables (except geographic coordinates) for the whole area: only for the individuals in the dataset. I am kind of lost...I have been searching for a couple of days now and I tried the predict.gam function with the easy type="response" and the more mysterious type="lpmatrix", and other possibilities but cannot find what I am looking for. I only calculate the risk for my individuals. I thought that the non-parametric spline component of the GAM/GAMM models could have helped me interpolate and "fill the gaps". Did I miss something big? Are there solut...

...rix%*%vcov(gam_fit,dispersion=1)%*%t(basis_matrix) stdv_f_x <- sqrt(diag(sigma_2_epsilon * smooth_matrix)) set.seed(123) R_sim <- 100000 mean_val <- rep(0, nrow(cov_beta_diff)) coef_diff <- mvrnorm(n = R_sim, mu = mean_val , Sigma = cov_beta_diff) Xp <- predict(gam_fit, type = "lpmatrix") sim_bet <- Xp %*% t(coef_diff) abs_ratio <- abs(sim_bet / stdv_f_x) ratio_max <- apply(abs_ratio, 2, max) cv <- quantile(ratio_max, probs = 0.95) So everything is fine until this point. My problem is:everything has been presented conditional on the smoothing parameter estimate...

Dear all, I fitted a non-parametric model using GAM function in R. i.e., gam(y~s(x1)+s(x2)) #where s() is the smooth function Then I obtained the coefficients(a and b) for the non-parametric terms. i.e., y=a*s(x1)+b*s(x2) Now if I want to use this estimated model to do optimization or sensitivity analysis, I am not sure how to incorporate the smooth function since s() may not

I fitted a GAM model with Poisson distribution using the function gam() in the mgcv package. My model is of the form: mod<-gam(y~s(x0)+s(x1)+s(x2),family=poisson). To extract estimates at a specified set of covariate values I used the gam `predict' method. But I want to get estimate and standard error of the difference of two fitted values. Can someone explain what should I do? Thank

Hi all, I am using GAM to model time trends in a logistic regression. Yet I would like to extract the the fitted spline from it to add it to another model, that cannot be fitted in GAM or GAMM. Thus I have 2 questions: 1) How can I fit a smoother over time so that I force one knot to be at a particular location while letting the model to find the other knots? 2) how can I extract the matrix

How can I obtain the values of confidence intervals from gam anf glm objects? Thanks in advance -- David Nogu?s Bravo Functional Ecology and Biodiversity Department Pyrenean Institute of Ecology Spanish Research Council Av. Monta?ana 1005 Zaragoza - CP 50059 976716030 - 976716019 (fax)

R-help, Sorry for posting again the same question (dated 26-03-2007) but all my mails have been sent to the recycle bin without possibility of recovering and thus I don't know if anyone has answer my query. Here is the original message: I'm applying a gam model (package mgcv) to predict relative abundances of a fish species. The covariates are year, month, vessel and statistical

Hello, Is it possible to estimate prediction interval using GAM? I looked through ?gam, ?predict.gam etc and the mgcv.pdf Simon Wood. I found it can calculate confidence interval but not clear if I can get it to calculate prediction interval. I read "Inference for GAMs is difficult and somewhat contentious." in Kuhnert and Venable An Introduction to R, and wondering why and if that