21th International Symposium on Forecasting, Callaway Gardens (United States of America). 17-20 June 2001
Summary:
In the new century, forecasting as another many tasks for managers is going to be characterised by the availabitlity of a lot of methods. Moreover, the forecast horizons are every time more diverse. On the one hand, forecast combining is an adequate methodology to handle the above scenario. On the other hand, such a setting is conceptually suitable to apply multivariate analysis. This paper reviews some of the main problems to combine efficiently forecast and studies the way to employ several methods of multivariate analysis in those problems. In particular, a methodology to elaborate combined forecasts with a large number of forecasts is proposed. Finally, several practical contexts are considered in order to assess the usefulness of such methodology.
Keywords: Combining forecasting, multivariate Analysis, time series, forecasting methodology, forecasts comparison, automated forecasting models
Publication date: 2001-06-17.
Citation:
C. Maté, A conceptual framework to combine forecasts based on classical multivariate analysis methods, 21th International Symposium on Forecasting, Callaway Gardens (United States of America). 17-20 June 2001.
