Question: Explain why a wildfire-detection model with 2% estimated bias and 9% estimated variance is described as having high variance, and compare its performance on the training set with its performance on the development set.

Sample answer: The model is described as having high variance because the estimated difference between its training error and development-set error is large: 9%. That means it fits the training data very closely, with only about 2% error there, but it does much worse on the development set, where the error rises to about 11%. This pattern shows that the model does not generalize well to new data. When a model performs much better on training data than on unseen data, it is overfitting.

Key points:

• Variance is estimated from the gap between training and development-set error, which is 9%.
• Training error is low, so bias is low at about 2%.
• Development-set error is much higher than training error.
• The model is overfitting rather than generalizing well.

Rubric: The response should identify the 9% training-dev gap as the variance estimate, note that the low training error indicates low bias, and explain that the larger development-set error shows overfitting and poor generalization.