Hello, in the Multivariate Exploratory ROC Analysis section of the biomarker analysis, it is stated that ROC curves were generated using Monte Carlo cross-validation (MCCV) with balanced sub-sampling, and that the procedure was repeated multiple times to estimate model performance and confidence intervals. Are we able to find out how many MCCV iterations were performed? Thanks!
I assume you refer to the default parameters used by the web server. Here is the R code used by web (from MetaboAnalystR):
if(nrow(data) > 500){
nRuns ← 10;
}else if(nrow(data) > 200){
nRuns ← 20;
}else if(nrow(data) > 100){
nRuns ← 30;
}else{
nRuns ← 50;
}
1 Like
To confirm, is nrow the number of features in the dataset? Thanks!
This topic was automatically closed after 2 days. New replies are no longer allowed.