In tune_penCSC:
Introduced tri.list to specify training data indices for arbitrary resampling.
Function now uses max(1L,ceiling(parallelly::availableCores()/2)) as the default number of cores for parallel computations.
In tune_penCSC:
Fixed an issue where the function only worked when the survival outcome included censoring.
grow.by has been deprecated and is scheduled for removal in a future version. The paths to reach maximum lambdas are now specified by glmnet.
Function now correctly sets multisession plan for parallel computations.
In penCSC:
In predict.penCSC:
newdata had to include all variables from the data when setting matching factor levels to characters in newdata.In tune_penCSC:
keep & standardize arguments were not affecting the $final_fits in the result.