Incorporating observed and unobserved heterogeneity in route choice analysis with sampled choice sets

An increasing number of researchers have shown an interest in enhancing discrete choice models by incorporating psychological and behavioral factors. The main objective of this paper is to explore the effect of observed and unobserved heterogeneity on route choice. The mixed logit framework is applied to consider the heterogeneity. In contrast with previous research, the repeat choice problem is dealt with by treating the random coefficients as consisting of three parts: individual specific term, O-D pair specific term and choice situation specific term. The solution of choice set generation in this study is the sampling method based on random walk. The sampling biases are corrected in the choice models. GPS data collected by private vehicles in Toyota city, Japan is used to estimate the choice models proposed in this study. This empirical analysis demonstrates that incorporation of observed characteristics and unobserved O-D pair specific heterogeneity can enhance route choice models significantly. It is also confirmed that drivers' taste is significantly affected by age, gender, vehicle displacement, O-D distance and familiarity with the O-D.