English

Empirical risk minimization is consistent with the mean absolute percentage error

Machine Learning 2015-09-09 v1

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.

Keywords

Cite

@article{arxiv.1509.02357,
  title  = {Empirical risk minimization is consistent with the mean absolute percentage error},
  author = {Arnaud De Myttenaere and Bénédicte Le Grand and Fabrice Rossi},
  journal= {arXiv preprint arXiv:1509.02357},
  year   = {2015}
}

Comments

in French, 47\`emes Journ\'ees de Statistique de la SFdS, Jun 2015, Lille, France. 2015

R2 v1 2026-06-22T10:51:45.203Z