Which metric is appropriate for evaluating a regression model's predictive accuracy?

• A. Mean Absolute Error
• B. F1 Score
• C. ROC-AUC
• D. Precision

Answer: (A)

In regression, you want to quantify how close your numerical predictions are to the actual values. Mean Absolute Error fits this goal because it averages the absolute differences between predicted and true values, giving an error measure in the same units as the target. This makes the result easy to interpret—on average, how far off are the predictions? MAE treats all errors equally and doesn’t square them, so it reflects typical error magnitude without letting a few large mistakes dominate the score.

The other metrics shown are for classification tasks: F1 Score balances precision and recall for binary outcomes, ROC-AUC assesses the model’s ability to rank positives over negatives, and Precision measures the proportion of correct positive predictions. These do not measure how close numerical predictions are to actual values, which is why MAE is the best choice for evaluating a regression model’s predictive accuracy.