It might be useful in general to allow mboost models with mstop=0, i.e., models that have been initialized with the optimal constant prediction but without any iterations.
For example as baseline for comparison.
Explicitly this is required to solve this issue for gamboostLSS boost-R/gamboostLSS#22
It might be useful in general to allow mboost models with
mstop=0, i.e., models that have been initialized with the optimal constant prediction but without any iterations.For example as baseline for comparison.
Explicitly this is required to solve this issue for gamboostLSS boost-R/gamboostLSS#22