Question: What is the main theoretical concern when you add new input features after doing error analysis, and what practical step can you take if that concern appears in experiments?

Sample answer: Adding more features can theoretically make the model more flexible and therefore increase variance. That means the model may fit the training data more closely but become less stable on new examples. If experiments show that the added features are helping but also raising the validation error gap, a practical fix is to use regularization to control complexity. In many cases, regularization offsets the extra variance so the new features can still improve performance overall.

Key points:

• Extra features can raise variance in theory.
• The practical response is regularization.
• Regularization helps control overfitting and can reduce the variance increase.

Rubric: A strong response will identify higher variance as the theoretical risk and name regularization as the practical method used to address it.