Abstract: This paper reports ongoing work carried out on an EPSRC-funded project investigating the potential of error components logit models (i.e. random coefficients logit, mixed logit or logit kernel models) for the study of drivers' route and departure time choices. Route and departure time choice represent a new applications for such models. This paper concentrates on the route choice decision, and consists of theoretical, analytical and empirical investigations. The paper discusses the limitations of the conventional multinomial logit (Luce, 1959; Domencich and McFadden, 1975) and multinomial probit (Thurstone, 1927) models for route choice applications, and considers a range of alternative models including C-logit (Cascetta et al., 1996), nested logit (Ben-Akiva, 1973, McFadden, 1978), paired combinatorial logit (Chu, 1981), cross-nested logit (Vovsha, 1997), generalised nested logit (Wen and Koppleman, 2000), heteroscedastic extreme value logit (Bhat, 1995) and error components logit (Cardell and Dunbar, 1980). The paper gives particular attention to error components logit, which offers a generalisation of, or an approximation to, a wide range of discrete choice models (e.g. McFadden and Train, 2000). Error components logit is able to represent complex patterns of correlation explicitly, hence its applicability to route choice. The paper reports an analytical comparison of several models for a simple route choice problem and an empirical comparison of several models using simulated route choice data from three networks of varying complexity. Estimation difficulties were encountered and overcome to a limited extent. Substantial differences were found between the models in their ability to reflect the network structure and to fit to simulated data. It is concluded that the error components logit model approach offers the best potential for progress but that other models of the logit family are also worth attention. (A) For the covering abstract see ITRD E121926.
Publication Year: 2001
Publication Date: 2001-01-01
Language: en
Type: article
Access and Citation
Cited By Count: 6
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot