similar to: different results with plot.lm vs. plot.lm(which=c(2))

I am running GLM models using the gamma family. For example: model <-glm(y ~ x, family=Gamma(link="identity")) I am getting different results for the normal Q-Q plot and the Scale-Location plot if I run the diagnostic plots without specifying the plot vs. specifying the plot ... e.g., "plot(model)" gives me a different Normal Q-Q graph than "plot(model,

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

I'm confused as to the trustworthiness of the dispersion parameters reported by glm. Any help or advice would be greatly appreciated. Context: I'm interested in using a fitted GLM to make some predictions. Along with the predicted values, I'd also like to have estimates of variance for each of those predictions. For a Gamma-family model, I believe this can be done as Var[y] =

Hello, I'm trying to use AIC to choose between 2 models with positive, continuous response variables and different error distributions (specifically a Gamma GLM with log link and a normal linear model for log(y)). I understand that in some cases it may not be possible (or necessary) to discriminate between these two distributions. However, for the normal linear model I noticed a discrepancy

Hello, I am looking for an explanation and/or fix for a discrepancy in the behaviour of the R lm.influence() function [ version R 1.5.0 (2002-04-29) ] and the same function in Splus [ Splus version 5.1 release 1, running on SGI IRIX 6.2]. The discrepancy is of concern because I am migrating some Splus scripts to R and need to ensure consistency of results. Specifically, when I fit a glm()

Dear all, I am using the bigglm package to fit a few GLM's to a large dataset (3 million rows, 6 columns). While trying to fit a Poisson GLM I noticed that the coefficient estimates were very different from what I obtained when estimating the model on a smaller dataset using glm(), I wrote a very basic toy example to compare the results of bigglm() against a glm() call. Consider the

Could I have some suggestions as to how (various ways) I can display my confidence interval results? rm(list = ls()) set.seed(1) func <- function(d,t,beta,lambda,alpha,p.gamma,delta,B){ d <- c(5,1,5,14,3,19,1,1,4,22) t <- c(94.32,15.72,62.88,125.76,5.24,31.44,1.048,1.048,2.096,10.48) post <- matrix(0, nrow = 11, ncol = B) theta <- c(lambda,beta) beta.hat <- 2.471546 for(j

Hi, I downloaded both gee and geepack, and I am trying to understand the differences between the two libraries. I used the same data and estimated the same model, with a correlation structure autoregressive of order 1. Surprisingly for me, I found very different results. Coefficients are slightly different in value but sometimes opposite in sign. Moreover, the estimate of rho (correlation

How can I get the results of lm() into a list so I can loop through the results? e.g. myResults[1] <- lm(...) myResults[2] <- lm(...) myResults[3] <- lm(...) ... myResults[15] <- lm(...) myResults[16] <- lm(...) so far every attempt I've tried doesn't work throwing a "number of items to replace is not a multiple of replacement length" error or simply not

Hello, I have found a problem with using the glm function with a gamma family. I have a vector of data, assumed to be generated by a gamma distribution. The parameters of this gamma distribution are estimated in two ways (i) using the glm() function, (ii) "by hand", using the optim() function. I find that the -2*likelihood at the maximum found by (i) is substantially larger than that

Dear R-Help I 'm using GLM function to Modelling. But when I used Gamma Family in GLM, then I can't run. It was error > glm(DamageRatio~MinTEMP+MaxTEMP+DayRain+Group1+Group2+Group3+Year,family=Gamma()) Error in eval(expr, envir, enclos) : Non-positive values not allowed for the gamma family Can Gamma Distribution use data begin 0 ? and then when I used GLM in S-Plus Program then

Hello all, I was wondering if someone can enlighten me as to the difference between the logLik in R vis-a-vis Stata for a GLM model with the gamma family. Stata calculates the loglikelihood of the model as (in R notation) some equivalent function of -1/scale * sum(Y/mu+log(mu)+(scale-1)*log(Y)+log(scale)+scale*lgamma(1/scale)) where scale (or dispersion) = 1, Y = the response variable, and mu

Hello, silly question I suppose, but somehow I can't manage to extract the probabilities from a glm.predict() result: > str(res) Named num [1:9] 0.00814 0.01877 0.025 0.02941 0.03563 ... - attr(*, "names")= chr [1:9] "1" "2" "3" "4" ... I got from: # A Gamma example, from McCullagh & Nelder (1989, pp. 300-2) clotting <-

Hi, Dear R-help I know there are some R package to deal with zero-inflated count data. But I am now looking for R package to deal with zero-inflated continuous data. The response variable (Y) in my dataset contains a larger mount of zero and the Non-zero response are quite right skewed. Now what i am doing is first to use a logistic regression on covariates (X) to estimate the probability of Y

Dear R users, Which test is most appropriate in glm when the family is Gamma? In the help page of anova.glm, I found the following ?For models with known dispersion (e.g., binomial and Poisson fits) the chi-squared test is most appropriate, and for those with dispersion estimated by moments (e.g., gaussian, quasibinomial and quasipoisson fits) the F test is most appropriate.? My questions :

I've downloaded R for windows (9.0.1) and it is great! I've converted all my lecture notes for my GLM course to run on R (they are available on my web page below). I must admit I particularly like the default contrast options, which are identical to GLIM. Also I like the gl function - very useful! I have a couple of questions/bugs: 1. predict.glm doesn't work, but predict.lm does -

Hi All, I am running a generalized linear model with gamma distribution in R (glm, family=gamma ) for my data (gene expression as response variable and few predictors). I want to calculate r-squared for this model. I have been reading online about it and found there are multiple formulas for calculating R2 (psuedo) for glm (in R) with gaussian (r2 from linear model), logistic regression

Dear R-List members, I want to check if a set of measurements follows better a gamma or a lognormal distribution (see data below). Using shapiro.test I can test for normality (shapiro.test(log (Lt)). To test for gamma (and normal) distribution I would use ks.test but I need to specify its shape and scale. How should I calculate these values in R? I tried > Lt.fit <- glm(Lt ~ 1,

Hello, I need to fit a generalized linear model with a Chi2 (6 ddl) as error distribution and with "log" as link function. I have looked in help(family) and maybe I could use Gamma(link="log") but I do not know if I can, and where I can define the shape and the scale arguments of the gamma distribution. Maybe there is another may to do that? Could someone explain me how can I

Quick question about the usage of glht. I'm working with a data set from an experiment where the response is bounded at 0 whose variance increases with the mean, and is continuous. A Gamma error distribution with a log link seemed like the logical choice, and so I've modeled it as such. However, when I use glht to look for differences between groups, I get significant