Adaptability of SVR Time Series Analysis Used in Forecasting of Logistics Demand

Forecast of logistics demand is a fundamental issue in research of logistics system. Generally, planning, management, and control of logistics system is involved in forecasting logistics demand. Common methods to forecast logistics demand are moving average, exponential smoothing, regression analysis etc. Support vector machines as originally introduced by Vapnik within the area of statistical learned theory and structural risk minimization have been proven to work successfully on many applications of nonlinear classification and estimation of function. Based on real operation data, the paper develops a time series analysis model with Support Vector Machine to forecast logistics demand. At last, different SVR model and traditional methods have been compared based on the index such as RME, RMSE. The adaptability of different SVR model in analysis of time series and forecasting logistics demand is illustrated in detail based on true scenario in practice. Final results indicate that, to some extent, SVR has some advantages in predicting logistics demand.