similar to: mu^2(1-mu)^2 variance function for GLM

This is not so much a bug as an infelicity in the code that can easily be fixed. The initialize expression in the quasi family function is, (uniformly for all links and all variance functions): initialize <- expression({ n <- rep.int(1, nobs) mustart <- y + 0.1 * (y == 0) }) This is inappropriate (and often fails) for variance function "mu(1-mu)".

In a glm() call using a quasi() family, one may define a custom variance function in the form of a "list containing components varfun, validmu, dev.resids, initialize and name" (quoting the help page for family). In trying to do so, I run into the following issue that I have not seen discussed previously: x <- runif(1000, min=0, max=1) y <- x + rnorm(1000, mean=0, sd=1)*x^(3/4)

Dear R users I have been looking for functions that can deal with overdispersed poisson models. Some (one) of the observations are negative. According to actuarial literature (England & Verall, Stochastic Claims Reserving in General Insurance , Institute of Actiuaries 2002) this can be handled through the use of quasi likelihoods instead of normal likelihoods. The presence of negatives is not

Dear R-Users, I wish to create a new family object based on the Binomial family. The only difference will be with the link function. Thus instead if using the 'logit(u)' link function, i plan to use '-log(i-u)'. So far, i have tried to write the function following that of the Binomial and Negative Binomial families. The major problem i have here is with the definition of the

How might you fit a generalized linear model (glm) with variance = mu+theta*mu^2 (where mu = mean of the exponential family random variable and theta is a parameter to be estimated)? This appears in Table 2.7 of Fahrmeir and Tutz (2001) Multivariate Statisticial Modeling Based on Generalized Linear Models, 2nd ed. (Springer, p. 60), where they compare "log-linear model fits to

G'day all, I had a look at the GLM code of R (1.4.1) and I believe that there are problems with the function "glm.fit" that may bite in rare circumstances. Note, I have no data set with which I ran into trouble. This report is solely based on having a look at the code. Below I append a listing of the glm.fit function as produced by my system. I have added line numbers so that I

Greetings, everyone. I've been trying to analyze bird nest survival data using generalized linear mixed models (because we documented several consecutive nesting attempts by the same individuals; i.e. repeated measures data) and have been unable to persuade the various GLMM models to work with my user-defined link function. Actually, glmmPQL seems to work, but as I want to evaluate a suite of

Dear R friends, When fitting GLMs in R, we may need to specify the variance function to do our analysis. I had thought it's the mean value, but it seems not. Could anybody expain the correct meaning of *mu* in the variance function of GLMs? The following content is from the R-hlep. variance for all families other than quasi, the variance function is determined by the family. The quasi

Hi, I am trying to run a generalized linear regression using a negative binomial error distribution. However, I want to use an overdispersion parameter that varies (dependent on the length of a stretch of road) so glm.nb will not do. >From what I've read I should be able to do this using GLM by specifying my own quasi family and describing the variance function using varfun, validmu,

I am resubmitting this to r-bugs, since Thomas Lumley indicates that it might be an error: On Wed, 5 Jul 2000, Thomas Lumley wrote: > On Wed, 5 Jul 2000, halvorsen wrote: > > > Hola! > > > > I have the following simple function: > > > > > testcar > > function(pow){ > > ob <- glm(Pound~CG+Age+Vage,data=car,weights=No, > >

under "Details" (version 2015-11-29 r69717; not quite cutting-edge, but nothing has changed in src/library/stats/man/family.Rd in 5 months [sorry for using the Github mirror, but I prefer the interface ... <https://github.com/wch/r-source/blob/trunk/src/library/stats/man/family.Rd>]) it says: valid.mu: logical function. Returns ?TRUE? if a mean vector ?mu? is within the

Hi all I am converting an S-Plus function into R. The S-Plus code uses some of the glm families, and family objects. The family objects in S-Plus and R have many different features, for example: In R: > names(Gamma()) [1] "family" "link" "linkfun" "linkinv" "variance" [6] "dev.resids" "aic"

I apologize for my earlier posting that, unbeknownst to me before, apparently was not in the correct format for this list. Hopefully this attempt will go through, and no-one will hold the newbie mistake against me. I could really use some help in writing a new glm link function in order to run an analysis of daily nest survival rates. I've struggled with this for weeks now, and can at least

Hi, I would like apply different link functions from Aranda-Ordaz (1981) family to large binary dataset (n = 2000). The existing links in glm for binomial data (logit, probit, cloglog) are not adequate for my data, and I need to test some other transformations. Is it possible to do this in R? And how? Thank you for your help, /Sharon

I know this has been dealt with before on this list, but the previous messages lacked detail, and I haven't figured it out yet. The model is: \x_{ij} = \mu + \alpha_i + \beta_j \alpha is a random effect (subjects), and \beta is a fixed effect (condition). I have a link function: p_{ij} = .5 + .5( 1 / (1 + exp{ -x_{ij} } ) ) Which is simply a logistic transformed to be between .5 and 1.

Greetings, everybody. Can I ask some glm questions? 1. How do you find out -2*lnL(saturated model)? In the output from glm, I find: Null deviance: which I think is -2[lnL(null) - lnL(saturated)] Residual deviance: -2[lnL(fitted) - lnL(saturated)] The Null model is the one that includes the constant only (plus offset if specified). Right? I can use the Null and Residual deviance to

Dear R users, I'm struggling with applying a user-defined link function in lmer. For analyzing data of a 2AFC psychophysical experiment, I would like to model my binary data with a logistic function with a lower limit at 0.5 instead of 0. In a previous question this has been described as a halflogit function. To do so I wrote my own link function and would like to submit it to lmer, however

Craig A Faulhaber wrote: >I am interested in using a generalized linear mixed model with data > that best fits a beta distribution (i.e., the data is bounded between > 0 and 1 but is not binomial). .. >For clarification, here's what I'm trying to model: >I have a beta-distributed response variable (y). I have a fixed-effect >explanatory variable (treatment),

Does anyone know what algorithm R uses in glm, family=binomial (i.e. a logit model)? I assume that it's in the source somewhere, but I wasn't able to find it. I'd like to know what file it's in (in a unix distribution of R). Thanks for your help. --------------------------- Barnet Wagman wagman at enteract.com 1361 N. Hoyne, 2nd floor Chicago, IL 60622 773-645-8369

Below is a SIMEX object that was generated with the "simex" function from the "simex" package applied to a logistic regression fit. From this mountain of information I would like to extract all of the values summarized in this line: .. ..$ variance.jackknife: num [1:5, 1:4] 1.684 1.144 0.85 0.624 0.519 ... Can someone suggest how to go about doing this? I can extract the