where y is the fitted freight volume, x1 is GDP, x2 is the fixed investment. Fitted freight volume versus actual freight volume based on MLR,NLR, and SLR is shown in Fig. 2. But in [7], the predicted transport data is compared with actual data. Finally the polynomial model gives the best result. Th emodel based on NLR is the polynomial model in this paper. So we can get the root mean squared error of calibration (RMSEC) of MLR,NLR,and SLR models from Table2: