Using the Mean Absolute Percentage Error for Regression Models
This work addresses the theoretical implications of using MAPE in regression for researchers and practitioners, but it is incremental as it builds on existing error metrics.
The paper investigates the use of Mean Absolute Percentage Error (MAPE) for regression models, showing that optimizing for MAPE is equivalent to weighted MAE regression and that Empirical Risk Minimization remains universally consistent with MAPE.
We study in this paper the consequences of using the Mean Absolute Percentage Error (MAPE) as a measure of quality for regression models. We show that finding the best model under the MAPE is equivalent to doing weighted Mean Absolute Error (MAE) regression. We show that universal consistency of Empirical Risk Minimization remains possible using the MAPE instead of the MAE.