Combining Predictive Schemes in Short-Term Traffic Forecasting

摘要

The principal motivation for combining forecasts which can either be a class lable (classification) or numerical (regression) has been to avoid the a priori choice of which forecasting method to use by attempting to aggregate all the information which each forecasting model embodies. In selecting the 'best' model, the forecaster is often discarding useful independent evidence in those models which are rejected. Hence the methodology of combining forecasts is founded upon the axiom of maximal information usage. Short-term traffic prediction is an area where the combining of two or more predictions is a promising technique which would directly improve the forecast accuracy. This approach may eventually help in specifying underlying processes more appropriately and thus build better individual models. This article deals with combining forecast methods potentially suitable for short-term prediction with their performance comparisons. The emphasis lies on the application to the short-term traffic flow prediction. Since the combination of predictors has, for the most part, implicitly assumed a stationary underlying process, attention has been focused on taking into account the effect of nonstationarity of the traffic flow process.