The dataset used to provide an unbiased evaluation of a final model fit on the training dataset. ‹ Supervised Learning up Text Analysis › Book