Question: In data mining the candidate model should be applied to a data set that was not used in the estimation process in order to find out the accuracy on unseen data; that unseen data set. What is the unseen data set called? How is the unseen data set selected?