ARMA-GRNN for passenger demand forecasting

摘要

Passenger transport demand analysis is strategically important in mastering ever-changing market for each transport mode. In order to improve the predict accuracy in complex reality situation, the ARMA-GRNN technique is proposed to capture both the linear and nonlinear perspectives of the intercity passenger demand forecast problem. Taking flight demand from 1991 to 2008 in Beijing-Shanghai corridor as an example, the numerical experiment results demonstrate that after subtract from the linear part by AR, the GRNN network based on principal component analysis can effectively fit the non-linear section with maximum error 1.08%.