The linear regression method and support vector machine method are used respectively to make a prediction on the cargo throughput of Lianyungang port. Based on the two methods, an optimal linear combined forecasting model is built. By contrasting the results of three methods, it indicates that combined forecasting method is more accurate and steady-going. In the end, the final prediction of cargo throughput of Lianyungang port from year 2010 to 2015 is given according to the combined forecasting model.