Study of SVM-Based Air-Cargo Demand Forecast Model

This paper analyzed some existing problems of the present air-cargo forecast methods. Then it established the SVM (support vector machine) model for air-cargo demand forecasting. Taking the historical statistical data of Beijing to Shanghai cargo volumes from Jan-2005 to Mar-2006 as fitting and forecasting specimens, we can obtain the prediction model to optimize, which was compared with that of Brown cubic exponential smoothing, by analyzing fitting and forecasting effect of model for different number of input nodes respectively. The result showed that the fitting effect by the model based on support vector machine was better than that of Brown. The former has a higher forecasting accuracy.