A New Phase Space Reconstruction Method for Prediction of Public Transit Passenger Volume

This article proposes a new method for prediction of public transit passenger volume based on phase space reconstruction and support vector machine (SVM); applies mutual information method in calculating the optimal time delay of time series of public transit passenger volume; applies Cao's method in calculating the optimal embedding dimension; then calculates out the Lyapunov exponent, evidencing that passenger volume chaos phenomena exists. It establishes phase space reconstruction - support vector machine prediction model as well as determines the pairs of training sample for prediction of the real passenger volume. It is proved through actual examples that this method can predict passenger volume effectively.