Check your understanding of this module before moving on.
?
0/3
Checkpoint quiz 3 questions
A quick check on this module's key ideas before you move on
Question 01
Which curve should you actually use to decide when to stop adding model complexity?
B is correct. Training error is not a trustworthy signal for choosing complexity, since it keeps improving even as the model starts overfitting. Repeatedly checking the test set during tuning would also invalidate it as an honest final check.
Question 02
What is the purpose of k-fold cross-validation, compared to a single train/validation split?
B is correct. Every sample gets to be in the training set for most rounds and the validation set for exactly one round, which uses the data more efficiently and averages out the luck of any single split.
Question 03
What is the clearest practical sign of overfitting?
B is correct. A widening gap between strong training performance and weaker validation performance is the textbook signature of overfitting, the model is starting to memorize rather than generalize.