ESTIMATION OF PARAMETERS ON NETWORK EQUILIBRIUM MODEL USING LINK FLOW DATA: MAXIMUM LIKELIHOOD METHOD CONSIDERING LINKS’ CORRELATION

Abstract

Link flow data are convenient for estimating parameters on network equilibrium models because they are easy to access. The minimum square error method has been adopted for parameter estimation in many cases. The minimum square error method presupposes that link flows are mutually independent, but actual link flows are not independent, especially, correlation between adjacent links is very high. Therefore, the minimum square error method is not necessarily appropriate from the statistical standpoint and the estimated parameters might be biased although we may be able to estimate parameters using it practically. In this study, we propose a method for estimating parameters on network equilibrium models which considers the correlation between links. Also, we examine which conditions are needed for estimating them consistently.