similar to: firth's penalized likelihood bias reduction approach

Dear R-Help List, I'm trying to implement Firth's (1993) bias correction for log-linear models. Firth (1993) states that such a correction can be implemented by supplementing the data with a function of h_i, the diagonals from the hat matrix, but doesn't provide further details. I can see that for a saturated log-linear model, h_i=1 for all i, hence one just adds 1/2 to each count,

Hi, I got two questions and would really appreciate any help from here. First, is the penalized maximum likelihood estimation(Firth Type Estimation) only fit for binary response (0,1 or TRUE, FALSE)? Can it be applied to multinomial logistic regression? If yes, what's the formula for LL and U(beta_i)? Can someone point me to the right reference? Second, when I used *logistf *on a dataset with

Dear all, I am encountering some issues with my data and need some help. I am trying to run glm analysis with a presence/absence variable as response variable and several explanatory variable (time, location, presence/absence data, abundance data). First I tried to use the glm() function, however I was having 2 warnings concerning glm.fit () : # 1: glm.fit: algorithm did not converge # 2:

Dear List, With regard to the question I previously raised, here is the result I obtained right now, brglm() does help, but there are two situations: 1) Classifiers with extremely high AIC (over 200), no perfect separation, coefficients converge. in this case, using brglm() does help! It stabilize the AIC, and the classification power is better. Code and output: (need to install package:

I''m running brglm with binomial loguistic regression. The perhaps multicollinearity-related feature(s) are: (1) the k IVs are all binary categorical, coded as 0 or 1; (2) each row of the IVs contains exactly C (< k) 1''s; (3) k IVs, there are n * k unique rows; (4) when brglm is run, at least 1 IV is reported as involving a singularity. I''ve tried recoding the n

Dear List, Apologies for this off-topic post but it is R-related in the sense that I am trying to understand what R is telling me with the data to hand. ROC curves have recently been used to determine a dissimilarity threshold for identifying whether two samples are from the same "type" or not. Given the bashing that ROC curves get whenever anyone asks about them on this list (and

Dear R list, From the lrm() binary logistic model we derived the G2 value or the likelihood-ratio chi-squared statistic given as L.R. model, in the output of the lrm(). How can this value be penalized for non-linearity (we used splines in the lrm function)? lrm.iRVI <- lrm(arson ~ rcs(iRVI,5), penalty=list(simple=10,nonlinear=100,nonlinear.interaction=4)) This didn’t work

What objective function is maximized by survreg with the default Weibull model? I'm getting finite parameters in a case that has the likelihood maximzed at Infinite, so it can't be a simple maximum likelihood. Consider the following: ############################# > set.seed(3) > Stress <- rep(1:3, each=3) > ch.life <- exp(9-3*Stress) > simLife <- rexp(9,

dear R-helper, I tried to find out a package in which i can have penalized maximum likelihood estimator applying on generalized extreme value distribution with beta function) but could not. would you please help me to know the name of the package. thanks for your help. S.Murshed --- r-help-request at stat.math.ethz.ch wrote: > Send R-help mailing list submissions to > r-help at

dear community, I am a beginer in R , and can't predict with logistic model in package logistf, could anyone help me ? thanks ! the following is my command and result : >library(logistf) >data(sex2) >fit<-logistf(case ~ age+oc+vic+vicl+vis+dia, data=sex2) >predict(fit,newdata=sex2) Error in predict(fit, newdata = sex2) : no applicable method for "predict"

I use smooth.spline(x, y) in package modreg and I would like to get value of penalized log likelihood and preferable also its two parts. To make clear what I am asking for (and make sure that I am asking for the right thing) I clarify my problem trying to use the same notation as in help(smooth.spline): I want to find the natural cubic spline f(x) such that L(f) = \sum_{k=1}{n} w[k](y[k] -

Hi everyone, I'm wondering how to select the "best" model when using logistf? AIC does not work neither does anova. I tried fitting a glm model but got the separation warning message so I tried using the logistf package but as I stepwise simplify the model I don't know if the simplification is motivated or not... Can anyone explain to me how I should approach this problem? I

Hi all, I have run a (glm) analysis where the dependent variable is the gender (family=binomial) and the predictors are percentages. I get a warning saying "fitted probabilities numerically 0 or 1 occurred" that is indicating that quasi-separation or separation is occurring. This makes sense given that one of these predictors have a very influential effect that is depending on a

Hi, all, Anyone has experience on Logistf package? I am using logistftest in Logistf package to selelct diagnosis genes. The result seems not the same as I expected. I have 10 gene expression data for 27 tumor 1 and 11 tumor 0. I want to select the best one using Maximum likelihood ratio test in logistic regression model. This is the way my code works: 1. Read in 10 genes as independent

I am new to using R. Currently, I am using the logistf package to run logistic regression analysis. When I run the following line of code: attach(snpriskdata) logisticpaper<-logistf(sascasecon~saspackyrs+newsbmi+EDUCATION+sasagedx+sasflung+condobst+sasadultasprev) I get the following error message: Error in sum(y) : invalid 'type' (character) of argument What does this error

First post/newbie post... bear with me. What I''m trying to achive (music site): A system containing tracks and moods with a HABTM relationship. I''ve got all that set up and functioning in the admin environment - i.e. admins can apply a variety of moods to a particular track via a series of checkboxes. Join table works just fine. I''m currently stuck on allowing

Dear list, when running a mood.test() (part of package "stats") on slightly longer vectors (than the example from the help-page) we get the error-message shown below : once both vectors tested are of length 50 this error oocurs. Note, that this problem didn't occur with R-2.7.x (or even older versions). > x <- rnorm(50,10,5) > y <- rnorm(50,2,5) > >

Hi there, I've got a set of 10 numeric variables called Mood1 to Mood10 in a dataset called mood. I'm trying to create a set of 10 new variables called m1 to m10 so that m1=Mood1*1, m2=Mood2*2, etc to m10=Mood10*10 Trawling through the internet, I eventually tried the following code: for (i in 1:10){ assign(x=paste0("mood$m",i),

Rather than go to a penalized GLM, you might be better off investigating the sources of quasi-perfect separation and simplifying the model to avoid or reduce it. In your data set you have several factors with large number of levels, making the data sparse for all their combinations. Like multicolinearity, near perfect separation is a data problem, and is often better solved by careful

UseRs, I have a dataframe with 2547 rows and several hundred columns in R 3.1.3. I am trying to run a small logistic regression with a subset of the data. know_fin ~ comp_grp2+age+gender+education+employment+income+ideol+home_lot+home+county > str(knowf3) 'data.frame': 2033 obs. of 18 variables: $ userid : Factor w/ 2542 levels