Which term describes a model with low bias but high variance?

Which term describes a model with low bias but high variance?

• A. Ridge Regression (L2 penalty)
• B. Overfitted models
• C. LASSO (L1 penalty)
• D. K-fold Cross-Validation

Answer: (B)

When a model is overfitted, it captures not only the underlying pattern but also the random noise in the training data. This makes the model fit the training set extremely well, giving it very low bias on that data. But because it’s focusing on noise rather than signal, its predictions become highly sensitive to the specifics of the training data, resulting in high variance across different data samples and poor generalization to new data. Methods like ridge regression and LASSO impose penalties to keep the model simpler, which tends to increase bias a bit but reduce variance, helping generalization. K-fold cross-validation is a technique for estimating how a model will perform on new data, not a type of model itself. So the term that best describes a model with low bias but high variance is overfitted models.

When a model is overfitted, it captures not only the underlying pattern but also the random noise in the training data. This makes the model fit the training set extremely well, giving it very low bias on that data. But because it’s focusing on noise rather than signal, its predictions become highly sensitive to the specifics of the training data, resulting in high variance across different data samples and poor generalization to new data. Methods like ridge regression and LASSO impose penalties to keep the model simpler, which tends to increase bias a bit but reduce variance, helping generalization. K-fold cross-validation is a technique for estimating how a model will perform on new data, not a type of model itself. So the term that best describes a model with low bias but high variance is overfitted models.