The Combination Forecast Model Of Port Throughput About Chongqing

As an important economic indicator of ports, port throughput shows theeconomic vitality of the port areas. Port throughput can accurately report period bysea into and out of the port areas and the quantity of loading and unloading of goods.It is the most important index to measure the size of the port, reflecting the port scaleand ability. Correct prediction of port throughput provides data for the planning anddevelopment of water transportation. It is also of great importance for determining areasonable and scientific port layout, the basic infrastructure investment, operationstrategy, development strategy and the related comprehensive transportation.According to the development of the logistics industry area, Port throughputprediction has the significant meaning, which also provides the basis for the planningof the regional logistics demand. In the contemporary environment with fastdevelopment, the affecting factors are complex which influencing the port throughput,including economic, political factors and logistics system of each link.Forecasting for the Chongqing port cargo throughput, firstly using single modelto predict the cargo throughput, selecting four kinds of forecasting methods, the greymodel GM(1,1)、three times exponential smoothing forecasting model, the trendforecasting method and unary linear regression model, established four differentforecasting methods. It is clear that the grey GM (1,1) is better than the other methodswith the lower error, meanwhile the three times exponential smoothing forecastingmodel prediction error is higher.In order to make more accurate prediction results, and effectively improve theprediction accuracy, this thesis introduces the use of combination forecast method toforecast. There also introduced Induction ordered weighted average (IOWA) operator,with the help of IOWA, establishing the IOWA combination forecast model, In theaccuracy of model test, selected five errors test indexes, included that squareerror(SSE), mean square error(MSE), mean absolute error percentage(MAEP), meanabsolute error(MAE), mean square error percentage(MSPE). Calculating the foursingle forecasting model of average prediction and the weighted average induction and orderly combined forecasting error, and finding that the error of orderlycombination forecast model were obviously lower than the others, which explains thatthe combination forecast model can improve the accuracy of prediction effectively.Although lots of the prediction model are predicted directly, but do not analysisthe stability of the model, this thesis make an analysis on the combination forecastmodel stability condition, and after the disturbance of the original data, thecomparative analysis shows that the stability of prediction model is more stable thanothers.