similar to: function to compare Brier scores from two models?

Hello, I am using the val.prob function in the Design package. I understand how the Brier quadratic error score is calculated, but I do not know how the Brier score computed on the calibrated rather than raw predicted probabilities (B cal) is calculated. My question is: how are the calibrated probabilities calculated? Any explanation of this, or references to explanations of this, would be

System: R 2.3.1 on a Windows XP computer. I am validating several cancer prognostic models that have been published with a large independent dataset. Some of the models report a probability of survival at a specified timepoint, usually at 5 and 10 years. Others report only the linear predictor of the Cox model. I have used Harrell's c index for censored data (rcorr.cens) as a measure of

Hello I've decided to try and distill an earlier rather ill focused question to try and elicit a response. Any help is greatly appreciated. Why does mod.cox not work with sbrier whilst mod.km does? Can I make it work? > data(DLBCL) > DLBCL.surv<-Surv(DLBCL$time,DLBCL$cens) > > mod.km<-survfit(DLBCL.surv) > mod.cox<-survfit(coxph(DLBCL.surv~IPI, data=DLBCL)) >

Dear all, I would like to obtain the Brier score prediction error at different times t for an extended Cox model. Previously I have used the 'pec' function (pec{pec}) to obtain prediction error curves for standard Cox PH models but now I have data in the counting process format (I have a covariate with a time-varying effect) and it seems that the pec function does not support the counting

Hi, I've searched online, in a few books, and in the archives, but haven't seen this. I believe that xerror is scaled to rel error on the first split. After fitting an rpart object, is it possible with a little math to determine the percentage of true classifications represented by a xerror value? -seth -- View this message in context:

Hi, I would like to specify a spherical correlation structure for spatially autocorrelated residuals in a model based upon the logistic function of a response that is a proportion (0 to 1) (so usual binary logistic regression is not an option). There is no need for a g-side random effect with grouping in this model. Am I correct that nlme requires this (meaning a correlated error structure only

Hi, Let's say I have a data frame (called "example") with numeric values stored (columns V1 and V2). I also have a string variable storing this name x1<-"example" Is there a way to use the variable x so that R knows that I want the specified action to occur on the data frame? For example, summary (x) would return a summary of the data frame? I am considering this

Dear, When fitting the following model knots <- 5 lrm.NDWI <- lrm(m.arson ~ rcs(NDWI,knots) I obtain the following result: Logistic Regression Model lrm(formula = m.arson ~ rcs(NDWI, knots)) Frequencies of Responses 0 1 666 35 Obs Max Deriv Model L.R. d.f. P C Dxy Gamma Tau-a R2 Brier 701 5e-07 34.49

Hi, I am reading a large space-delimited text file into R (41 columns and many rows) and need to do run each row's values through another R object and then write to another text file. So, far using readLines and writeLines seems to be the best bet. I've gotten the data exchange working except each row is read in as one 'chunk', meaning the row has all values between two quotes

Dear List, I am familier with binary models, however i am now trying to get predictions from a ordinal model and have a question. I have a data set made up of 12 categorical predictors, the response variable is classed as 1,2,3,4,5,6, this relates to threat level of the species ( on the IUCN rating). Previously i have combined levels 1 and 2 to form = non threatened and then combined 3-6 to

Dear List, I am looking to run a host of models (60) with three methods - lmer,glm and lrm. Is there a way to output the key stats into a table that i can copy to excel? I.e for lmer i would want AIC,BIC etc for lrm i would want Brier score, r2, c-value etc At present i am running the models from a script and then copying across the values into a excel spreadsheet however this is time

Hi, I am attempting to evaluate the prediction error of a coxph model that was built after feature selection with glmnet. In the preprocessing stage I used na.omit (dataset) to remove NAs. I reconstructed all my factor variables into binary variables with dummies (using model.matrix) I then used glmnet lasso to fit a cox model and select the best performing features. Then I fit a coxph model

Hi List, I don't understand why 'lrm' doesn't recognize the '~.' formula. I'm pretty sure it was working before. Please see below: I'm using R2.3.0, WinXP, Design 2.0-12 thanks, ...Tao > dat <- data.frame(y=factor(rep(1:2,each=50)), x1=rnorm(100), x2=rnorm(100), x3=rnorm(100)) > lrm(y~., data=dat, x=T, y=T) Error in terms.formula(formula, specials =

Dear All, I have just estimated this model: ----------------------------------------------------------- Logistic Regression Model lrm(formula = Y ~ X16, x = T, y = T) Model Likelihood Discrimination Rank Discrim. Ratio Test Indexes Indexes Obs 82 LR chi2 5.58 R2 0.088 C 0.607 0

Hi, I've come across a strange error when using the lrm.fit function and the subsequent predict function. The model is created very quickly and can be verified by printing it on the console. Everything looks good. (In fact, the performance measures are rather nice.) Then, I want to use the model to predict some values. I get the following error: "fit was not created by a Design

Hi, I am trying to run a simple logistic regression using lrm() to calculate a odds ratio. I found a confusing output when I use summary() on the fit object which gave some OR that is totally different from simply taking exp(coefficient), see below: > dat<-read.table("dat.txt",sep='\t',header=T,row.names=NULL) > d<-datadist(dat) > options(datadist='d')

Dear Sir or Madam? I am a doctor of urology,and I am engaged in developing a nomogram of bladder cancer. May I ask for your help on below issue? I set up a dataset which include 317 cases. I got the Binary Logistic Regression model by SPSS.And then I try to reconstruct the model ?lrm(RECU~Complication+T.Num+T.Grade+Year+TS)? by R-Project,and try to internal validate the model through

You must create a output filter. Edit /etc/printcap and add your client printer like this: ... PC-PRINTER:\ :sd=/var/spool/lpd/PP01:\ :mx#0:\ :lp=/dev/null:\ :if=/var/spool/lpd/Filter/PP01:\ :sh: ... Adjust the printcap parameters as needed (spool directory, directory to find the filter script). Make sure the lpd can access the directory. Consult your manual. Now create the filter

from samba 3.0.9 to 3.0.13 I can not access my shares from a NT4 box as usual before upgrading. I can access samba shares from linux via mount //server/testuser /mnt -o username=testuser,password=testpass as before. Even the newest version (3.0.14a) from samba.org doesnt't help. What has happened to samba ??? Any hints ?? Thanks for your thoughts, G?tz.

I've noticed that glm and lrm give extremely different results if you attempt to fit a saturated model to a dataset with zero cells. Consider, for instance the data from, Agresti's Death Penalty example [0]. The crosstab table is: , , PENALTY = NO VIC DEF BLACK WHITE BLACK 97 52 WHITE 9 132 , , PENALTY = YES VIC DEF BLACK WHITE BLACK 6 11