Abstract:We introduce the basic principle of the curve fitting degree analysismethod in the article, and expoundthe role that the method is applied to regression forecast. By selecting dataof freight and GDP in Hangzhouduring "EleventhFive-Year" period, we construct several regression models and analyze theircurve fitting degrees, and we optimize the model that is selected. By examining,predictions of the obtained model are relatively close to the actual trend.
Keywords：Curve fitting; Traffic;Turnover; Regression forecast
1. Introduction
Forecastinghighway freight volume is an important part for traffic planning. The competentauthority will take the freight as one of basis for project approval,decision-making, determining construction scale and so on. The highway freightvolume, including freight volume and freight turnover, is a reflection of quantitativeindex that transportation industry affected on national economy. And it is alsoan important index to study how fast the market economy developed. Highwayfreight volume is the actual tonnage of goods transported by road in a certainperiod of time. And highway freight turnover is a composite unit（ton-km）, which is consisted of weight and distance of delivered goods, incertain period. The index of highway freight turnover includes not only thequantity of goods, but also the factor of distance, so this index is more comprehensiveto reflect the transportation production.
In recently domestic andinternational freight volume forecasting researches, we usually adopt a varietyof forms such as combination model, unbiased grey forecasting model, RBF neuralnetwork, regression curve model. Because the regression curve model is a kindof specific, effective and highly practical value prediction method, it isoften adopted. Due to the transportation system is a sub-system of socialeconomic system, the development of social economy determines the development oftransportation demand. On the other hand, the rapid development of transportationindustry will also prompt the development of social economic system. Theyconnect closely. From the analysis of recent years date, such as GDP, freightvolume and freight turnover, we will find out strong positive correlation,which basically comply with the conditions of regression forecasting. But thereare various regression forecasting models, and results predicted by differentmodels are quite different from actual volume in the accuracy or reliability.