Which metric computes the average magnitude of errors, treating all errors equally?

Which metric computes the average magnitude of errors, treating all errors equally?

• A. MAE
• B. MSE
• C. RMSE
• D. MAPE

Answer: (A)

Mean Absolute Error measures the average size of the prediction errors without regard to direction. It does this by taking the absolute value of each residual (the difference between predicted and actual), then averaging those magnitudes. Because every error is treated by its magnitude alone, all errors contribute equally to the final score. This differs from metrics that square errors, like MSE or RMSE, which give more weight to larger mistakes and thus emphasize big errors more. It also differs from MAPE, which expresses errors as percentages of the actual values, introducing scale effects and potential issues when actual values are small or near zero. For example, if errors are 2 and 20 across two observations, MAE is (|2| + |20|)/2 = 11, reflecting the average magnitude of errors without bias toward the direction or the size of any particular error beyond its absolute amount.

Mean Absolute Error measures the average size of the prediction errors without regard to direction. It does this by taking the absolute value of each residual (the difference between predicted and actual), then averaging those magnitudes. Because every error is treated by its magnitude alone, all errors contribute equally to the final score.

This differs from metrics that square errors, like MSE or RMSE, which give more weight to larger mistakes and thus emphasize big errors more. It also differs from MAPE, which expresses errors as percentages of the actual values, introducing scale effects and potential issues when actual values are small or near zero.

For example, if errors are 2 and 20 across two observations, MAE is (|2| + |20|)/2 = 11, reflecting the average magnitude of errors without bias toward the direction or the size of any particular error beyond its absolute amount.