The use of prediction error to optimize the number of splitting rules in a tree model does not control the probability of the emergence of splitting rules with
Clinical trials are usually long term studies and it seems impossible to reach all required subjects at the same time. Performing interim analyses and monitorin