Performance Evaluation of Forecasting Models : At Last, Problem Solved!

この論文をさがす

抄録

The current methodology for assessing the relative performance of competing forecasting models is unidimensional in nature in that models are compared to each other using a single criterion at a time, which typically leads to different rankings for different criteria thus resulting in conflicting results or conclusions regarding the performance of competing models. Recently, Xu and Ouenniche (2011, 2012a, 2012b) raised and addressed this methodological issue by proposing several data envelopment analysis (DEA) and multi-criteria decision making analysis (MCDA) frameworks for determining a single ranking that takes account of several criteria. Very recently Ouenniche, Xu and Tone (2013, 2014) built on the previous work by Xu and Ouenniche and proposed improved DEA-based models to address this novel application of DEA. While this above mentioned work has focused on the relative performance evaluation of forecasting models of continuous variables such as the levels and the volatilities of prices with application to crude oil prices, Mousavi, Ouenniche and Xu (2014a, 2014b) focused on the relative performance evaluation of forecasting models of discrete variables with application to bankruptcy prediction and used both DEA and MCDA methodologies to perform a comparative analysis of bankruptcy prediction models under multiple criteria. In all applications examined so far, empirical results suggest that multi-criteria frameworks for comparing the relative performance of forecasting models provide valuable tools to apprehend the true nature of the relative performance of forecasting models.

収録刊行物

詳細情報

  • CRID
    1571980077856973184
  • NII論文ID
    110009830775
  • NII書誌ID
    AN00351206
  • 本文言語コード
    en
  • データソース種別
    • CiNii Articles

問題の指摘

ページトップへ