similar to: `power' documentation slip (PR#8838)

Full_Name: simon wood Version: 2.2.0 (and lower) OS: linux/windows Submission from: (NULL) (86.135.153.59) I think that the documentation for the `specials' attribute of a `terms.object' is not quite right: specials: If the 'specials' argument was given to 'terms.formula' there is a 'specials' attribute, a list of vectors indicating the terms

Hello, I have a question regarding model selection and dropping of terms for GAMs fitted with package mgcv. I am following the approach suggested in Wood (2001), Wood and Augustin (2002). I fitted a saturated model, and I find from the plots that for two of the covariates, 1. The confidence interval includes 0 almost everywhere 2. The degrees of freedom are NOT close to 1 3. The partial

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

For summary(GAM) in the mgcv package smooth the degrees of freedom for the F value for test of smooth terms are the rank of covariance matrix of \hat{beta} and the residuals df. I've noticed that in a lot of GAMs I've fit the rank of the covariance turns out to be 9. In Simon Wood's book, the rank of covariance matrix is usually either 9 or 99 (pages 239-230 and 259). Can anyone

R-users E-mail: r-help@r-project.org My understanding is that package "mgcv" is based on "Generalized Additive Models: An Introduction with R (by Simon N. Wood)". On the page 126 of this book, eq(3.4) looks a quartic equation with respect to "x", not a cubic equation. I am wondering if all routines which uses cubic splines in mgcv are based on this quartic

Dear list, I do apologize if these are basic questions. I am fitting some GAM models using the mgcv package and following the model selection criteria proposed by Wood and Augustin (2002, Ecol. Model. 157, p. 157-177). One criterion to decide if a term should be dropped from a model is if the estimated degrees of freedom (EDF) for the term are close to their lower limit. What would be the

Dear listers, I'm using gam(from mgcv) for semi-parametric regression on small and noisy datasets(10 to 200 observations), and facing a problem of overfitting. According to the book(Simon N. Wood / Generalized Additive Models: An Introduction with R), it is suggested to avoid overfitting by inflating the effective degrees of freedom in GCV evaluation with increased "gamma"

Hi, Currently the `gaussian' family's initialization code signals an error if any response data are zero or negative and a log link is used. Given that zero or negative response data are perfectly legitimate under the GLM fitted using `gaussian("log")', this seems a bit unsatisfactory. Might it be worth changing it? The current offending code from `gaussian' is:

x <- NA na.act <- na.action(na.exclude(x)) y <- rep(0,0) naresid(na.act,y) ... currently produces the result... numeric(0) ... whereas the documentation might lead you to expect NA The behaviour is caused by the line if (length(x) == 0L) return(x) in `stats:::naresid.exclude'. Removing this line results in the behaviour I'd expected in the above example (and in a

Hi all R users ! I'm using gamm function from Simon Wood's mgcv package, to fit a spatial regression Generalized Additive Mixed Model, as covariates I have the geographical longitude and latitude locations of indexed data. I include a random effect for each district (dist) so the code is fit <- gamm(y~s(lon,lat,bs="tp", m=2)+offset(log(exp.)), random=list(dist=~1),

Dear all, I am using the "mgcv" package by Simon Wood to estimate an additive mixed model in which I assume normal distribution for the residuals. I would like to test this model vs a standard parametric mixed model, such as the ones which are possible to estimate with "lme". Since the smoothing splines can be written as random effects, is it correct to use an (approximate)

Dear Mr. Simon Wood, dear list members, I am trying to fit a similar model with gam from mgcv compared to what I did with BayesX, and have discovered the relatively new possibility of incorporating user-defined matrices for quadratic penalties on parametric terms using the "paraPen" argument. This was really a very good idea! However, I would like to constraint the coefficients

Hello, I am a bit confused about gamm in mgcv. Consulting Wood (2006) or Ruppert et al. (2003) hasn't taken away my confusion. In this code from the gamm help file: b2<-gamm(y~s(x0)+s(x1)+s(x2)+s(x3),family=poisson,random=list(fac=~1)) Am I correct in assuming that we have a random intercept here....but that the amount of smoothing is also changing per level of the

Hello, I have been trying to find out whether it is possible to use more than one offset in a gam (in mgcv). The reason I would like to do this is to 1) account for area surveyed in a Poisson model of sightings of porpoises within defined grid cells (each cell has a slightly different area) and 2) account for detection probability within each grid cell (some grid cells are further away from the

In Dr. Wood's book on GAM, he suggests in section 4.1.6 that it might be useful to shrink a single smooth by adding S=S+epsilon*I to the penalty matrix S. The context was the need to be able to shrink the term to zero if appropriate. I'd like to do this in order to shrink the coefficients towards zero (irrespective of the penalty for "wiggliness") - but not necessarily all the

i read a book from WOOD, there's an example which is talking about the pollutant. library(gamair) library(mgcv) y<-gam(death~s(time,bs="cr",k=200)+s(pm10median,bs="cr")+s(so2median,bs="cr")+s(o3median,bs="cr")+s(tmpd,bs="cr"),data=chicago,family=Possion) lag.sum<-function(a,10,11) {n<-length(a) b<-rep(0,n-11) for(i in 0:(11-10))

We are using GAM in mgcv (Wood), relatively new users, and wonder if anyone can advise us on a problem we are encountering as we analyze many short time series datasets. For each dataset, we have four models, each with intercept, predictor x (trend), z (treatment), and int (interaction between x and z). Our models are Model 1: gama1.1 <- gam(y~x+z+int, family=quasipoisson) ##no smooths Model

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

Thanks Simon, by cloning a smooth construct do you mean copying and modifying the smooth constructor code? Could you pleas elaborate on your answer? Which is the Predict.matrix method? 2018-01-12 3:20 GMT-06:00 Simon Wood <simon.wood at bath.edu>: > There probably is a way, but it involves some programming. You would need to > clone a smooth constructor (e.g. for the "cr"

Hello everyone,          Dr. wood told me that I can adapting his example to force cubic spline to pass through certain point.          I still have no idea how to achieve this. Suppose we want to force the cubic spline to pass (1,1), how can I achieve this by adapting the following code? # Penalized example: monotonic penalized regression spline ..... # Generate data from a monotonic truth.