Research work · 2015

Empirical risk minimization is consistent with the mean absolute percentage error

Arnaud De Myttenaere, Bénédicte Le Grand, Fabrice Rossi

arXiv (Cornell University)

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 also show that, under some asumptions, universal consistency of Empirical Risk Minimization remains possible using the MAPE.

Download PDF

DOI: 10.48550/arxiv.1509.02357