prediction for ranking groups

Hi, I'd like to range my sample by some intervals and have the result as mearge of of intervals for all predictors and the rising of my result in every interval was more then in previus 5%. could you help me?

Take a look at caret::createDataPartitions It will divide your sample into equal intervals. Create a function to run your model on a list containing each of the resulting sets to create a list of estimators.

That said unless you are using the cumulative results (i.e., increasing sample size), there's no reason to expect that the estimator isn't going to vary randomly.

Bootstrap sampling may be something to explore as an alternative.

1 Like

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.