similar to: Consistency of Logistic Regression

Christoph Lehman had problems with seperated data in two-class logistic regression. One useful little trick is to penalize the logistic regression using a quadratic penalty on the coefficients. I am sure there are functions in the R contributed libraries to do this; otherwise it is easy to achieve via IRLS using ridge regressions. Then even though the data are separated, the penalized

Hi, First let me say I am a big fan of R and appreciate all your time and effort. The update() function does not seem to work in a for loop. Consider the following: mdat <- matrix(c(1,2,3, 11,23,13, 12,4,8), nrow = 3, ncol=3, byrow=TRUE) reg <- lm(mdat[7:9]~1) for(i in 1:2) { reg <- update(reg,.~.+mdat[((i-1)*3 + 1):(i*3)]) #update reg twice } reg # reg should have two independent

I'm using FC4 and R 2.3.1 to fit a mixed effects logistic regression. The response is 0/1 and both the response and the age are the same for each pair of observations for each subject (some observations are not paired). For example: id response age 1 0 30 1 0 30 2 1 55 2 1 55 3 0 37 4 1 52 5 0 39 5 0 39 etc. I get the

Hi all, I am also confused about the manual: a. The input arguments: wt.method are the weights case weights (giving the relative importance of case, so a weight of 2 means there are two of these) or the inverse of the variances, so a weight of two means this error is half as variable? w (optional) initial down-weighting for each case. init (optional) initial values for the

Hi, In the help files in the ?mgcv package for the gam.control() function, there is an option irls.reg. The help files describe this option as: For most models this should be 0. The iteratively re-weighted least squares method by which GAMs are fitted can fail to converge in some circumstances. For example, data with many zeroes can cause problems in a model with a log link, because a mean of

I have written a program to solve a particular logistic regression problem using IRLS. In one step, I need to integrate something out of the linear predictor. The way I'm doing it now is within a loop and it is as you would expect slow to process, especially inside an iterative algorithm. I'm hoping there is a way this can be vectorized, but I have not found it so far. The portion of code

Dear list, I'm trying to fit a glm model using family=poisson(link = "identity"). The problem is that the glm function fits a model with a negative intercept, which sounds like a nonsense to me, being the response a Poisson variable. >From a previous discussion on this list I've understood that the glm function uses IRLS for the fitting without any constraint so it is

Hi: I'm using function lmer from package lme4, and I get this message: " There were 12 warnings (use warnings() to see them)" So I checked them: Warnings 1 to 11 said: 1: optim returned message ERROR: ABNORMAL_TERMINATION_IN_LNSRCH in: "LMEoptimize<-"(`*tmp*`, value = structure(list(maxIter = 50, ... and Warning 12 said: 12: IRLS iterations for glmm did

Dear All, I have the following error message when I fitted lmer to a binary data with the "AGQ" option: Error in family$mu.eta(eta) : NAs are not allowed in subscripted assignments In addition: Warning message: IRLS iterations for PQL did not converge Any help? Thanks in advance, Abderrahim [[alternative HTML version deleted]]

Hi all! My question is why, and what I can do about that I sometimes, but not always, get warning-messages like nlminb returned message singular convergence (7) in: LMEopt(x = mer, value = cv) or IRLS iterations for PQL did not converge when trying to fit a model looking like this: lmer<-(cbind(Diseased,Healthy)~Fungus+(1|Family)+(1|Fungus:Family), family="binomial") to four

I'd like to estimate an ordinal logistic regression with a random effect for a grouping variable. I do not find a pre-packaged algorithm for this. I've found methods glmmML (package: glmmML) and lmer (package: lme4) both work fine with dichotomous dependent variables. I'd like a model similar to polr (package: MASS) or lrm (package: Design) that allows random effects. I was

Dear R users, I am trying to fit a GLMM to the following dataset; tab a b c 1 1 0.6 199320100313 2 1 0.8 199427100412 3 1 0.8 199427202112 4 1 0.2 199428100611 5 1 1.0 199428101011 6 1 0.8 199428101111 7 0 0.8 199527103011 8 1 0.6 199527200711 9 0 0.8 199527202411 10 0 0.6 199529100412 11 1 0.2 199626201111 12 2 0.8 199627200612 13 1 0.4 199628100111 14 1 0.8

Dear list members, In re-running with GLMM() from the lme4 package a generalized-linear mixed model that I had previously fit with glmmPQL() from MASS, I'm getting a warning of a convergence failure, even when I set the method argument of GLMM() to "PQL": > bang.mod.1 <- glmmPQL(contraception ~ as.factor(children) + cage + urban, + random=~as.factor(children) + cage +

I am trying to do a power analysis to get the number of replicas per treatment. If I try to get the power it works just fine: setn=c(2,3) sdx=c(1.19,4.35) power.t.test(n = setn, delta = 13.5, sd = sdx, sig.level = 0.05,power = NULL) If I go the other way to obtain the "n" I have problems. sdx=c(1.19,4.35) pow=c(.8,.8) power.t.test(n = NULL, delta = 13.5, sd = sdx, sig.level = 0.05,

Here's a bug one of my students noticed. When you call plot() on a glm object, plot.lm gets called. The second plot it shows is supposed to give a normal QQ plot of the standard deviance residuals, but it doesn't. The glm object created by glm.fit returns something (the IRLS weights?) in fit$weights which plot.lm takes as observation weights, so you get strange residuals in the QQ

Dear useRs, upon request, the VGAM package (currently version 0.7-1) has been officially released on CRAN (the package has been at my website http://www.stat.auckland.ac.nz/~yee/VGAM for a number of years now). VGAM implements a general framework for several classes of regression models using iteratively reweighted least squares (IRLS). The key ideas are Fisher scoring, generalized linear and

I am working on predicitng the scores for a days worth of matches of team sports. I have already collected data for the teams for the season we are concentrating on. I have been fitting poisson models for football games and have worked out what model is best and which predictor variables are most important. We would now like to predict the probability distribution for the scores for each team.

Sometimes the p.value returned by t.test() is the same that I calculate using pt() and sometimes it's not. I don't understand the difference. I'm sure there is a simple explanation but I haven't been able to find it, even after looking at the code for t.test.default. I apologize if this is a basic and obvious question. For example: > data(sleep) >

Hi Paul, Take a look at gam() from package mgcv (gam = generalized additive models), maybe this will help you. GAMs can work with other distributions as well. Generalized additive models consist of a random component, an additive component, and a link function relating these two components. The response Y, the random component, is assumed to have a density in the exponential family. I am not sure

Hello, I would like to know if there is any package that allow us to fit Generalized Linear Models via Maximum Likelihood and Linear Models using Generalized Least Squarse in R as the functions glim and gls, respectively, from S-Plus. Also, anybody know if there is any package that fit Log-Linear Models using Generalized Least Squares? Any help will be very useful. Thanks, -- Frederico