Improved Support Vector Regression and Its Application to Prediction of Railway Passenger Traffic Volume

To improve the prediction abilities of the present methods for railway passenger traffic volume,support vector regression(SVR)was improved by weighting penalty coefficients using Mahalanobis distance between training and testing samples,and a model for predicting the time serial of railway passenger traffic volume was set up based on the improved SVR.The prediction of railway passenger traffic volume from 1980 to 1998 shows that the proposed method can obtain a more accurate result than the BP artificial neural network.