similar to: Fwd: mgcv 1.4 on CRAN

mgcv 1.5-0 is now on CRAN. Main changes are: * REML and ML smoothness selection are now available. * A Tweedie family has been added. * `gam.method' has been replaced (see arguments `method' and `optimizer' for `gam') For other changes see the changeLog. Simon -- > Simon Wood, Mathematical Sciences, University of Bath, Bath, BA2 7AY UK > +44 1225 386603

re: Cubic splines in package "mgcv" I don't have access to Gu (2002) but clearly the function R(x,z) defined on p126 of Simon Wood's book is piecewise quartic, not piecewise cubic. Like Kunio Takezawa (below) I was puzzled by the word "cubic" on p126. As Simon Wood writes, this basis is not actually used by mgcv when specifying bs="cr". Maybe the point is

mgcv 1.5-0 is now on CRAN. Main changes are: * REML and ML smoothness selection are now available. * A Tweedie family has been added. * `gam.method' has been replaced (see arguments `method' and `optimizer' for `gam') For other changes see the changeLog. Simon -- > Simon Wood, Mathematical Sciences, University of Bath, Bath, BA2 7AY UK > +44 1225 386603

The plots look reasonable to me. The plot of residuals against linear predictor always looks scary when many of the fitted values are very close to zero, so I tend to look at residuals against sqrt(fitted) in such cases. I don't think that the presence of the zero curve is a reason to reject the model --- it's easy to produce such plots by fitting a completely correct model to simulated

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, According to the help file for model.frame/get_all_vars, the following should produce the same output from both functions, but it doesn't... > dat <- list(X=matrix(1:15,5,3),z=26:30) > model.frame(~z+X,dat) z X.1 X.2 X.3 1 26 1 6 11 2 27 2 7 12 3 28 3 8 13 4 29 4 9 14 5 30 5 10 15 > get_all_vars(~z+X,dat) [1] z X <NA> <NA> <0

Hi Greg, For quasi families I've used extended quasi-likelihood (see Mccullagh and Nelder, Generalized Linear Models 2nd ed, section 9.6) in place of the likelihood/quasi-likelihood in the expression for the (RE)ML score. I hadn't realised that this was possible before the paper was published. best, Simon ps. sorry for slow reply, the original message slipped through my filter for

Hi, I am using GAMs (package mgcv) to smooth event rates in a penalized regression setting and I was wondering if/how one can select the order of the derivative penalty. For my particular problem the order of the penalty (parameter "m" inside the "s" terms of the formula argument) appears to have a larger effect on the AIC/deviance of the estimated model than the

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

The help pages of R-1.2.2 include several pages on various time series functions, but when I try to use these functions they appear not to be available .... am I missing something obvious, or are these functions not yet built? Chris Rogers ----------------------------------------------------------------------- L C G Rogers, Professor of Probability tel:+44 1225 826224 Department of

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

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.

hi Using a ts or tprs basis, I expected gcv to decrease when increasing the basis dimension, as I thought this would minimise gcv over a larger subspace. But gcv increased. Here's an example. thanks for any comments. greg #simulate some data set.seed(0) x1<-runif(500) x2<-rnorm(500) x3<-rpois(500,3) d<-runif(500) linp<--1+x1+0.5*x2+0.3*exp(-2*d)*sin(10*d)*x3

I've got a data frame which I've split by a factor, creating a list of dataframes which I have then done various operations on individually. I next want to recombine the resulting dataframes (still held in a list, still with the same number of columns with the same names) and there does not appear to be a `good' way to do this - at the moment, I'm using a for-loop with the rbind

Hello All, I have a working port of R on my SL5500. I've not tested the X windowing support yet, but was more concerned about the accuracy of the fp emulation. The following is the result of the test which Stuart Leask recommended I should try: Mandrake 8.2 > x<-NA > is.na(x) [1] TRUE > x+1 [1] NA > 2*x [1] NA Zaurus OZ3 > x<-NA > is.na(x) [1] TRUE > x+1 [1] 1

Dear R users I have a list, as follows: > intvl.period.myrs $Devonian [1] 4.8 4.2 9.5 5.7 $Ordovician [1] 7.2 5.1 10.2 1.9 $Silurian [1] 4.7 3.0 7.8 2.0 3.3 1.6 2.6 2.7 I want to write a loop that will sum up the values in each part, and give me a vector containing the (in this case 3) summed values this is what I have so far: for (i in 1:length(names(intvl.periods.myrs)) {

Hello All, The link to the binary & installation instructions (tar.gz binary not an ipk I'm afraid) is as follows: http://students.bath.ac.uk/enpsgp/Zaurus/#R It eventually dawned on me that the WORDS_BIGENDIAN define (or lack thereof) was causing the problems (after testing ieee NaN compliance that is). When cross-compiling it's probably fair enough that the configure script

Hello All, I read in the archives that someone managed to compile R but that there were problems with the fp performance, specifically the handling of NaNs. Could someone please explain the problem to me and how to test whether it is occurring. I have just compiled R for the Zaurus 5500, and want to see whether it works or requires tweaking. Regards, Simon

Dear useRs, I am using mgcv version 1.7-16. When I create a model with a few non-linear terms and a random intercept for (in my case) country using s(Country,bs="re"), the representative line in my model (i.e. approximate significance of smooth terms) for the random intercept reads: edf Ref.df F p-value s(Country) 36.127 58.551 0.644