Data Science Model Evaluation and Validation 4

Question 1

High training accuracy but low test accuracy is a classic symptom of:

Accepted Answer

Overfitting

Answer

Underfitting

Answer

High bias

Answer

Data leakage prevention

Question 2

In the bias-variance tradeoff, a model with high bias typically:

Accepted Answer

Underfits and performs poorly on both training and test data

Answer

Overfits the training data

Answer

Has very low training error

Answer

Is too complex for the data

Question 3

Which technique most directly reduces overfitting by penalizing large coefficients?

Accepted Answer

Regularization (L1/L2)

Answer

Increasing the number of epochs

Answer

Removing the validation set

Answer

Using a larger learning rate

Question 4

Early stopping prevents overfitting during training by:

Accepted Answer

Halting training when validation performance stops improving

Answer

Removing features with low variance

Answer

Reducing the batch size to one

Answer

Increasing model depth

Question 5

A learning curve where both training and validation errors remain high suggests:

Accepted Answer

Underfitting (high bias)

Answer

Overfitting (high variance)

Answer

Perfect generalization

Answer

Data leakage

Question 6

Data leakage in model validation typically causes:

Accepted Answer

Overly optimistic validation scores that collapse in production

Answer

Underfitting on the training set

Answer

Increased training time only

Answer

Lower precision but higher recall

Question 7

Increasing model complexity generally has what effect on the bias-variance tradeoff?

Accepted Answer

Decreases bias but increases variance

Answer

Decreases both bias and variance

Answer

Increases bias and decreases variance

Answer

Has no effect on either