Reference:The overall classification accuracy was about 96.4% for the training data (902 out of 936) and was about 96.3% for the test data (387 out of 402).

Question:

Which accuracy would be better to report for these data, the training or test data, and why?

Options:

Training Data; This data is larger and should be more representative

Training Data; These data better mimic how the model would be used in the real world.

Test Data; These data were withheld and represent the most difficult cases to predict.

Test Data; These data were not used in the fitting process and represent new cases to evaluate