Re: [R] Package for .632 (and .632+) bootstrap and the cross-validation of ROC Parameters
spime wrote: Suppose I have Training data: my.train Testing data: my.test The bootstrap does not need split samples. I want to calculate bootstrap error rate for logistic model. My wrapper function for prediction pred.glm - function(object, newdata) { ret - as.factor(ifelse(predict.glm(object, newdata, type='response') 0.4, 0, 1)) return(ret) } But i thing i cant understand if i want to calculate misclassification error for my testing data what will be in my data in the following formula. Misclassification error has many problems because it is not a proper scoring rule, i.e., it is optimized by bogus models. Frank errorest(RES ~., data=???, model=glm, estimator=boot, predict=pred.glm, est.para=control.errorest(nboot = 10)) Using my.test got following error, Error in predict(mymodel, newdata = outbootdata) : unused argument(s) (newdata = list(RES = c(1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0), CAT01 = c(4, 4, 2, 4, 4, 4, 4, 4, 4, 2, 1, 2, 2, 4, 4, 4, 1, 1, 2, 2, 1, 4, 1, 4, 1, 4, 2, 4, 1, 4, 2, 3, 1, 1, 3, 3, 4, 2, 4, 2, 1, 2, 2, 1, 1, please reply... Frank E Harrell Jr wrote: spime wrote: Hi users, I need to calculate .632 (and .632+) bootstrap and the cross-validation of area under curve (AUC) to compare my models. Is there any package for the same. I know about 'ipred' and using it i can calculate misclassification errors. Please help. It's urgent. See the validate* functions in the Design package. Note that some simulations (see http://biostat.mc.vanderbilt.edu/rms) indicate that the advantages of .632 and .632+ over the ordinary bootstrap are highly dependent on the choice of the accuracy measure being validated. The bootstrap variants seem to have advantages mainly if an improper, inefficient, discontinuous scoring rule such as the percent classified correct is used. -- Frank E Harrell Jr Professor and Chair School of Medicine Department of Biostatistics Vanderbilt University __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. -- Frank E Harrell Jr Professor and Chair School of Medicine Department of Biostatistics Vanderbilt University __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Package for .632 (and .632+) bootstrap and the cross-validation of ROC Parameters
Suppose I have Training data: my.train Testing data: my.test I want to calculate bootstrap error rate for logistic model. My wrapper function for prediction pred.glm - function(object, newdata) { ret - as.factor(ifelse(predict.glm(object, newdata, type='response') 0.4, 0, 1)) return(ret) } But i thing i cant understand if i want to calculate misclassification error for my testing data what will be in my data in the following formula. errorest(RES ~., data=???, model=glm, estimator=boot, predict=pred.glm, est.para=control.errorest(nboot = 10)) Using my.test got following error, Error in predict(mymodel, newdata = outbootdata) : unused argument(s) (newdata = list(RES = c(1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1, 0), CAT01 = c(4, 4, 2, 4, 4, 4, 4, 4, 4, 2, 1, 2, 2, 4, 4, 4, 1, 1, 2, 2, 1, 4, 1, 4, 1, 4, 2, 4, 1, 4, 2, 3, 1, 1, 3, 3, 4, 2, 4, 2, 1, 2, 2, 1, 1, please reply... Frank E Harrell Jr wrote: spime wrote: Hi users, I need to calculate .632 (and .632+) bootstrap and the cross-validation of area under curve (AUC) to compare my models. Is there any package for the same. I know about 'ipred' and using it i can calculate misclassification errors. Please help. It's urgent. See the validate* functions in the Design package. Note that some simulations (see http://biostat.mc.vanderbilt.edu/rms) indicate that the advantages of .632 and .632+ over the ordinary bootstrap are highly dependent on the choice of the accuracy measure being validated. The bootstrap variants seem to have advantages mainly if an improper, inefficient, discontinuous scoring rule such as the percent classified correct is used. -- Frank E Harrell Jr Professor and Chair School of Medicine Department of Biostatistics Vanderbilt University __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code. -- View this message in context: http://www.nabble.com/Package-for-.632-%28and-.632%2B%29-bootstrap-and-the-cross-validation-of-ROC-Parameters-tf4068544.html#a11578129 Sent from the R help mailing list archive at Nabble.com. __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
[R] Package for .632 (and .632+) bootstrap and the cross-validation of ROC Parameters
Hi users, I need to calculate .632 (and .632+) bootstrap and the cross-validation of area under curve (AUC) to compare my models. Is there any package for the same. I know about 'ipred' and using it i can calculate misclassification errors. Please help. It's urgent. -- View this message in context: http://www.nabble.com/Package-for-.632-%28and-.632%2B%29-bootstrap-and-the-cross-validation-of-ROC-Parameters-tf4068544.html#a11561405 Sent from the R help mailing list archive at Nabble.com. __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
Re: [R] Package for .632 (and .632+) bootstrap and the cross-validation of ROC Parameters
spime wrote: Hi users, I need to calculate .632 (and .632+) bootstrap and the cross-validation of area under curve (AUC) to compare my models. Is there any package for the same. I know about 'ipred' and using it i can calculate misclassification errors. Please help. It's urgent. See the validate* functions in the Design package. Note that some simulations (see http://biostat.mc.vanderbilt.edu/rms) indicate that the advantages of .632 and .632+ over the ordinary bootstrap are highly dependent on the choice of the accuracy measure being validated. The bootstrap variants seem to have advantages mainly if an improper, inefficient, discontinuous scoring rule such as the percent classified correct is used. -- Frank E Harrell Jr Professor and Chair School of Medicine Department of Biostatistics Vanderbilt University __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.