similar to: Cross-validation for parameter selection (glm/logit)

Hello All; I need to run a multiple regression analysis and use Akaike's Information Criterion for model selection. I understand that this command will give the AIC value for specified models: AIC(object, ..., k = 2) with "..." meaning any other optional models for which I would like AIC values. But, how can I specify (in the place of "...") that I want R to

Hello, i am a newby on R and i am trying to make a backward selection on a binomial-logit glm on a large dataset (69000 lines for 145 predictors). After 3 days working, the stepAIC function did not terminate. I do not know if that is normal but i would like to try computing a "homemade" backward with a repeated glm ; at each step, the predictor with the max pvalue would be

It might help if you provided the code you used. It's possible that you didn't use direction="backward" in stepAIC(). Or if you did, it was still running, so whatever else you try will still be slow. The statement "R provides only the pvalues for each level" is wrong: look at the anova() function. Bob On 29 June 2017 at 11:13, Beno?t PELE <benoit.pele at

Hello, Are there any packages/functions available for testing the assumptions underlying assumptions for a good GLM model? Like linktest in STATA and smilar. If not, could somebody please describe their work process when they check the validity of a logit/probit model? Regards, Jay

The following is a simple example with a poor solution that shows my difficulties with scope. The function /logit.test /has 3 arguments: /model.start,/ an initial model; /model.finish/, an all-inclusive model, /my.data/, a dataset, in this case trivial. There are 2 function calls in l/ogit.test,/ first to /glm/ to get an initial fit (local variable /logit/) using /model.start;/ then a call

Dear R users: I have run a logistic regression, used Gelman et al.'s car package to simulate the parameter estimates of that model, and have plotted the probability (using Gelman et al.'s invlogit() function) of the dependent variable being 1 given the value of a particular independent variable is at its mean. The plot has probabilities on the y-axis and the number (1-1000) of the

This is a question at the border between stats and r. When I do a glm with many potential effects, and select a model using stepAIC, many independent variables are selected even if there are no relationship between dependent variable and the effects (all are random numbers). Do someone has a solution to prevent this effect ? Is it related to Bonferoni correction ? Is there is a ratio of

Problem: I am sorting through model selection process for first time and want to make sure that I have used glm, stepAIC, and update correctly. Something is strange because I get a different result between: 1) a glm of 12 predictor variables followed by a stepAIC where all interactions are considered and then an update to remove one specific interaction. vs. 2) entering all the terms

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.

Dear all, I am using glm function in order to estimate a logit model i.e. glm(Y ~ data[,2] + data[,3], family = binomial(link = "logit")). I also created a function that estimates logit model and I would like it to compare it with the glm function. So, does anyone know what optimizer or optimization method glm uses in order to derive the result? Thank you Dimitris -- View this

Posted about this earlier. Didn't receive any response But, some further research leads me to believe that MAYBE a GLMM or a GEE function will do what I need. Hello, I have a bit of a tricky puzzle with trying to implement a logit model as described in a paper. The particular paper is on horseracing and they explain a model that is a logit trained "per race", yet somehow the

Hi All, When modeling with glm and family = binomial (link = logit) and response values of 0 and 1, I get the predicted probabilities of assigning to my class one, then I would like to compare it with my vector y which does have the original labels. How should I change the probabilities into values of zero and 1 and then compare it with my vector y to find out about the accuracy of my

I want to predict a set of proportions from a linear regression using glm. The model includes a logit link. However I want to extract the variance:covariance matrix among the predicted proportions rather than on the logit scale. Is there a way to do this using Vcov or a similar package in R? Thanks Chris

Hi, Does anyone know of a function in R that can handle a fractional variable as the dependent variable? The catch is that the function has to be inclusive of 0 and 1, which betareg() does not. It seems like GLM might be able to handle the fractional logit model, but I can't figure it out. How do you format GLM to do so? Best, Rachael [[alternative HTML version deleted]]

If AIC is giving you a model that is too large, then use BIC (log(n) as the penalty for adding a term in the model). This will yield a more parsimonious model. Now, if you ask me which is the better option, I have to refer you to the huge literature on model selection. Best, Ravi [[alternative HTML version deleted]]

Dear useR's, I was comparing results for a logistic regression model between different library's. themodel formula is arranged as follows: response ~ (intercept) + value + group OR: glm( response ~ (intercept) + value + group , family=binomial(link='logit')) lrm( response ~ (intercept) + value + group ) ROC( from = response ~ (intercept) + value + group ,

Hi all, I am using glm function with family binomial(logit) to fit logistic regression model. My data is very big and the algorithm is such that it has to run glm function hundreds of times. Now *I need only the **estimates of the coefficients and std. error in my output, *but apparently glm function is computing several other statistics and parameters (mentioned below) which increases the

Hi I am trying to make glm() work to analyze a toy logit system. I have a dataframe with x and y independent variables. I have L=1+x-y (ie coefficients 1,1,-1) then if I have a logit relation with L=log(p/(1-p)), p=1/(1+exp(L)). If I interpret "p" as the probability of success in a Bernouilli trial, and I can observe the result (0 for "no", 1 for

Dear all, I would like to Boostrap the stepAIC() procedure from package MASS for variety of model objects, i.e., fn <- function(object, data, B = 2){ n <- nrow(data) res <- vector(mode = "list", length = B) index <- sample(n, n * B, replace = TRUE) dim(index) <- c(n, B) for (i in 1:B) { up.obj <- update(object, data = data[index[, i], ])

# 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