Selecting and ranking time series models using the NOEMON approach

In this work, we proposed to use the NOEMON approach to rank and select time series models. Given a time series, the NOEMON approach provides a ranking of the candidate models to forecast that series, by combining the outputs of different learners. The best ranked models are then returned as the selected ones. In order to evaluate the proposed solution, we implemented a prototype that used MLP neural networks as the learners. Our experiments using this prototype revealed encouraging results.