Processing Variable Forms for Travel Estimation

As applied to intercity travel demand estimation, the conventional gravity model consists of some pairing-like variables which attempt to express certain interactions affecting the travel potential between two cities. This paper explores alternative functional forms of the pairing-like variables, in order to produce a more effective travel demand forecast relationship. The study suggests the use of three statistical procedures for formulating the demand model, based on functional forms of variables that most significantly express the travel demand relationship. The best functional form of the pairing-like variable is defined as the one which will yield a statistically sound model of travel estimation, as indicated by stepwise multiple regression techniques. It is anticipated that by incorporating these variable forms, the predictive power of resulting travel demand models will be substantially improved, while data requirements are minimized.