search for: mustart

...d using `gaussian("log")', this seems a bit unsatisfactory. Might it be worth changing it? The current offending code from `gaussian' is: initialize = expression({ n <- rep.int(1, nobs) if (is.null(etastart) && is.null(start) && is.null(mustart) && ((family$link == "inverse" && any(y == 0)) || (family$link == "log" && any(y <= 0)))) stop( "cannot find valid starting values: please specify some") mustart <- y })...

Hello, Suppose one is interested in fitting a GLM with a log link to binomial data. How does R choose starting values for the estimation procedure? Assuming I don't supply them. Thanks, Tyler

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)". Here is a short demo to show it: ################################################# set.seed(666) dat <- data.frame(x = rep((-10):10, each = 5), w = rep(1:5, 21)) dat <- trans...

# Hello R Users, # # I would like to fit a glm model with quasi family and # logistical link function, but this does not seam to work # with binary data. # # Please don't suggest to use the quasibinomial family. This # works out, but when applied to the true data, the # variance function does not seams to be # appropriate. # # I couldn't see in the # theory why this does not work. # Is

...m (obtained from CRAN). (My processor is an AMD Athlon 64 x2 4800+). The error and traceback are *** caught segfault *** address 0x8001326f2b, cause 'memory not mapped' Traceback: 1: any(is.na(varmu)) 2: glm.fit(x = X, y = Y, weights = weights, start = start, etastart = etastart, mustart = mustart, offset = offset, family = family, control = control, intercept = attr(mt, "intercept") > 0) 3: glm(y ~ x, family = binomial("logit"), data = da) 4: ... Note that this does NOT happen every time I make this call to glm(). I'm calling glm() in a loop an...

...link=logit), trace = TRUE) Deviance = 28.34255 Iterations - 1 Deviance = 27.72554 Iterations - 2 Deviance = 27.72527 Iterations - 3 Warning message: non-integer #successes in a binomial glm! in: eval(expr, envir, enclos). But trying to exclude intercept (or pass anything to glm.fit, especially mustart to improve convergence if I have understand the role of mustart correctly): glm(as.matrix(y) ~ as.matrix(X), family=binomial(link=logit), intercept = FALSE, TRACE = TRUE) seems not to be a correct since it returns: Error in glm.control(...) : unused argument(s) (intercept ...) ### Same apply to...

Hello R Users, I am having difficulty deleting the last warning message in a loop so that the only warning that is produced is that from the most recent line of code. I have tried options(warn=1), rm(last.warning), and resetting the last.warning using something like: > warning("Resetting warning message") This problem has been addressed in a previous listserve string,

...~ x, family=poisson, data=df) ------ output --------------- *** caught segfault *** address (nil), cause 'unknown' Traceback: 1: ifelse(y == 0, 1, y/mu) 2: dev.resids(y, mu, weights) 3: glm.fit(x = X, y = Y, weights = weights, start = start, etastart = etastart, mustart = mustart, offset = offset, family = family, control = control, intercept = attr(mt, "intercept") > 0) 4: glm(y ~ x, family = poisson, data = df) -------------------------------- The code generates a segfault if the value of 'N' is greater than 16384. regards Adrian Ba...

...at's all for now. Thanks for the great package and I hope that this bug report will be useful. Cheers, Berwin ========================= original glm.fit ========================= 1 glm.fit <- 2 function (x, y, weights = rep(1, nobs), start = NULL, etastart = NULL, 3 mustart = NULL, offset = rep(0, nobs), family = gaussian(), 4 control = glm.control(), intercept = TRUE) 5 { 6 x <- as.matrix(x) 7 xnames <- dimnames(x)[[2]] 8 ynames <- names(y) 9 conv <- FALSE 10 nobs <- NROW(y) 11 nvars <- NCOL(x) 12...

I'm fitting some small data sets using a binomial glm with complementary log-log link. In some ill-conditioned cases I am getting convergence failures. I know how to adjust maxit and epsilon, but that doesn't seem to help. In fact I know some very good starting values for my fits, but I can't see how to get them in to glm(). How may I do this? -- Dr Murray Jorgensen

...with MCMClogit function (in MCMCpack package). For example, in case of glm function as given below, there is weights option in the arguments. Aparently there is no option of using weights in MCMClogit. glm(formula, family = gaussian, data, weights, subset, na.action, start = NULL, etastart, mustart, offset, control = glm.control(...), model = TRUE, method = "glm.fit", x = FALSE, y = TRUE, contrasts = NULL, ...) MCMClogit(formula, data = parent.frame(), burnin = 1000, mcmc = 10000, thin=1, tune=1.1, verbose = 0, seed = NA, beta.start = NA, b0 = 0, B0 = 0, user.prior....

...ently I'm missing something here...any help is much appreciated. Thanks, Leo. "gam2.slist" <- c("s", "lo", "random") "gam2" <- function(formula, family = gaussian, data, weights, subset, na.action, start = NULL, etastart, mustart, control = gam.control(...), model = FALSE, method="glm.fit", x = FALSE, y = TRUE, ...) { call <- match.call() mf <- match.call(expand.dots = FALSE) m <- match(c("formula", "data", "subset", "weights", "na.action"...

Hi all, New user here - I include the following command in the prompt read.csv("document.csv", header = TRUE ) and the output shows up. But when I include the following command summary(data) I get the following message "Error in object[[i]] : object of type 'closure' is not subsettable" Can someone please advise why R is not reading my data? Thanks

Hi, I am trying to run two Non-Gaussian regressions: logistic and probit. I am receiving two different errors when I try to run these regressions and I am not sure what they mean or how to fix my syntax. Here is the logistic regression error: Error in family$linkfun(mustart) : Argument mu must be a nonempty numeric vector Here is the probit regression error: Error in pmax(eta, -thresh) : cannot mix 0-length vectors with others The dataset that I am using has some missing data. R puts NA values in place of the missing values. I am not sure if this is what is caus...

Details in documentation: "All of ?weights?, ?subset?, ?offset?, ?etastart? and ?mustart? are evaluated in the same way as variables in ?formula?, that is first in ?data? and then in the environment of ?formula?." In fact, `data` is usually not an environment, and I have not seen arguments evaluated in `environment(formula)` when `data` is provided. (Information in `environment(fo...

...y), round(m), mu, log = TRUE)) } initialize <- expression({ if (NCOL(y) == 1) { if (is.factor(y)) y <- y != levels(y)[1] n <- rep.int(1, nobs) if (any(y < 0 | y > 1)) stop("y values must be 0 <= y <= 1") mustart <- (weights * y + 0.5)/(weights + 1) m <- weights * y if (any(abs(m - round(m)) > 0.001)) warning("non-integer successes in a binomial glm!") } else if (NCOL(y) == 2) { if (any(abs(y - round(y)) > 0.001))...

...I have not seen discussed previously: x <- runif(1000, min=0, max=1) y <- x + rnorm(1000, mean=0, sd=1)*x^(3/4) vf <- function(mu) { abs(mu)^(3/4) } vm <- function(mu) { rep(TRUE, length(mu)) } dr <- function(y, mu, wt) { (y-mu)^2 } it <- expression({ n <- rep.int(1, nobs); mustart <- y }) glm(y ~ x, family=quasi(link="identity", variance=list(varfun=vf, validmu=vm, dev.resids=dr, initialize=it, name="custom"))) This gives "Error in switch(vtemp, constant = { : EXPR must be a length 1 vector" from line 576 in file family.R (https://github.com...

...with 75 presence and 75 absence if instead I had randomly sampled 3150 background points I could set the weights to 1 for presence and 75/3150 for the background points. I simulate this example by repeating my 75 absences 42 times and setting the weights for the absence to 75/3150 and setting mustart equal to the predicted response from the 75/75 model and fitting glms to each data set. The output agrees in the coefficient estimates null and residual deviance but NOT in the AIC or the value returned from logLik() it appears the AIC is calculated from the likelihood though as I understand i...

...2 * sum(ifelse(m > 0, (wt/m), 0) * dbinom(round(m * y), round(m), mu, log = TRUE)) } initialize<- expression({ n<-rep(1,nobs) if (any(y < 0 | y > 1)) stop("y values must be 0 <= y <= 1") mustart <- (weights * y + 0.5)/(weights + 1) m <- weights * y if (any(abs(m - round(m)) > 0.001)) warning("non-integer #successes in a binomial glm!") }) environment(variance) <- environment(validmu) <- environment(dev.resids) <- envi...

..." is invalid: possible values are \"mu(1-mu)\", \"mu^2(1-mu)^2\", \"mu\", \"mu^2\", \"mu^3\" and \"constant\"", variancetemp), domain = NA)) initialize <- expression({ n <- rep.int(1, nobs) mustart <- y + 0.1 * (y == 0) }) aic <- function(y, n, mu, wt, dev) NA structure(list(family = "quasi", link = linktemp, linkfun = stats$linkfun, linkinv = stats$linkinv, variance = variance, dev.resids = dev.resids, aic = aic, mu.eta = stats$mu.eta, initi...