Suppose you train a classifier and test it on a held-out validation set. It gets 30% classification accuracy on the training set and 30% classification accuracy on the validation set.

a. From what problem is your model most likely suffering: underfitting or overfitting?

b. What could reasonably be expected to improve your classifier’s performance on the validation set: adding new features or removing some features? Justify your answer. item What could reasonably be expected to improve your classifier’s performance on the validation set: collecting more training data or throwing out some training data? Justify your answer.