Prediction Research Of Container Throughput Based On SVM

With the development of economy, the important of port is increasingly prominent, port is the interface of national foreign trade, it is the power for the development of the city. With the improvement of goods containerization rate, the scale of container transportation in the transport enlarges unceasingly, the port's container throughput directly affect the port layout planning of container throughput, so the effective and accurate prediction is necessary of formulating correct port development policy.Based on the analysis of container transportation development problems and mechanism in Guangzhou port, the paper summarized the influence factors of the container throughput, then combined entropy weights method and the fuzzy classification of the index classification, to find more important factors of affecting container throughput, then build predicting model of throughput based on support vector machine. Take Guangzhou port for example, predict its container throughput of next five years, forecasting result close to the container throughput aim for 2000 million which is in the "Guangzhou port development opinions".This paper's main achievements include:1. analysis factors affecting container throughput in the past, combining with the actual research, painted container throughput generation system feedback chart and combined with Guangzhou port container throughput established throughput factors affecting factors index system; 2. First try in combining the entropy value method with fuzzy classification to choose important index for combination; 3. Establish the support vector machine forecasting model.Through analysis of Guangzhou port container throughput forecast, it proofs SVM model in this paper have certain feasibility in practice, but accuracy is not very high, in this aspect still needs further research.