1st International Symposium on Interval Data Modelling: Theory and Applications - SIDM 2015, Pekín (China). 03-04 julio 2015
Resumen:
The new world arising from the global crisis, which began in December 2007, is more complex, uncertain and riskier than the world before 2008. In this scenario having accurate forecasts in economics, finance, energy, health and so on; is more critical than ever.
After more than 45 years of research, there is a general consensus "combining forecasts reduces the final forecasting error". As part of this consensus there is also a well-known fact "a simple average of several forecasts often outperforms complicated weighting schemes", which was named "forecast combination puzzle".
The introduction of interval time series (ITS) concepts and forecasting methods has been proposed in different papers. Given that we are able to use several forecasting methods with ITS, one main issue in symbolic data analysis with interval-valued data is how to combine several forecasts obtained for an ITS or obtained for a magnitude by a judgmental process developed with the participation of several companies, organizations or persons, such as is usual in the Delphi method, for example.
This talk review briefly the main combination schemes with classical time series (CTS) forecasts and propose several combination schemes with interval-valued time series (ITS) forecasts. The forecast combination puzzle in the ITS forecasts framework will be analyzed in the context of different accuracy measures. As a consequence an agenda for future research in the field of combining forecasts obtained for symbolic data will be outlined.
Palabras clave: forecast combination, interval forecast, interval-valued data, interval-valued information systems, optimal weight, symbolic data analysis.
Fecha de publicación: 2015-07-03.
Cita:
C. Maté, Combining interval time series forecasts. An agenda for future research, 1st International Symposium on Interval Data Modelling: Theory and Applications - SIDM 2015, Pekín (China). 03-04 julio 2015.
