Using the Mean Absolute Percentage Error for Regression Models

TL;DR

This paper explores the implications of using MAPE as a regression quality measure, revealing its equivalence to weighted MAE and its compatibility with consistent empirical risk minimization.

Contribution

It demonstrates that optimizing MAPE is equivalent to weighted MAE regression and maintains universal consistency in empirical risk minimization.

Findings

MAPE optimization is equivalent to weighted MAE regression

Universal consistency of ERM is preserved with MAPE

Insights into MAPE's suitability for regression evaluation

Abstract

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.