0
$\begingroup$

I would like to know how to do calibration plot with Hosmer-lemeshow test and AUC for ROC curve after multiple imputation in R. I build one prediction model and tried to do model performance but noticed that it is not simple.

I also tried pool_performance function in psfmi package, but showed the following error.

Error in cut.default(pred, quantile(pred, c(seq(0, 1, 1/groups_cal)))) : 
  'breaks' are not unique

Could you tell me if this function is ok to use for model performance with multiply imputed data set, otherwise the better R code and example?

$\endgroup$

1 Answer 1

0
$\begingroup$

You have to define a smaller number of groups using the groups_cal option in the function, e.g. try 8 or 6, because the spread of predicted probabilities of your model seems too small for 10 groups.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.