Evaluation of discrete choice and neural network approaches for modelling driver compliance with traffic information

Abstract

This article evaluates dynamic driver behaviour models that can be used, in the context of intelligent transport systems (ITS), to predict drivers' compliance with traffic information. The inputs to this type of models comprise drivers' individual socio-economic characteristics and other variables that may influence their compliance behaviour. The output is a binary integer representing whether drivers comply with travel advice or not. Two approaches are available for formulating this category of classification problems: discrete choice models and artificial neural networks (ANNs). The literature on this topic clearly points to the limitations of the discrete choice approach which suffers from assumptions of perfect information about travel conditions, infinite information processing capabilities of drivers and inability to model the uncertainty in driver decision making or the vagueness in information received from ITS devices. ANNs, on the other hand, are able to deal with complex non-linear relationships, are fault tolerant in producing acceptable results under imperfect inputs and are suitable for modelling reactive behaviour which is often described using rules, linking a perceived situation with appropriate action. This study aims to evaluate the performance of these two categories of models based on a common data set of driver behaviour, collected from a field behavioural survey on a congested commuting corridor in Brisbane, Australia. This article proposes the combination of fuzzy logic and neural networks as a viable approach for overcoming the limitations of existing algorithms by modelling drivers as heterogeneous individuals. The results showed superior performance for a neuro-fuzzy model over binary choice models in terms of classifying or predicting the categories of drivers most likely to comply (or not comply) with traffic advice. The accuracy of the proposed model, in terms of classification rate, ranged between 95 and 97% compared to 50-73% for the discrete choice models.

An Analysis of Car and Motorcycle Ownership in Macao

“…These models based on utility theory assume a utility function with observed variables and a random component capturing the unobserved factors that affect the individual's choice, and the random component may follow some assumed distributions. Dia and Panwai (2010) argued that assumptions of perfect information and decision-making capability of individuals in the discrete choice model may not be satisfied in some problems, and the complex relationship of imperfect input should be better modelled with other approaches. Their paper proposed an artificial neural network (ANN) model, and showed that ANN is superior to, in terms of classification rate, the discrete-choice models for the prediction problem of driver compliance with traffic information.…”

Effectiveness of en route traffic information in developing countries using conventional discrete choice and neural‐network models

“…A number of studies have employed NN models largely to model the diversion under the influence of en route traffic information. The history of such studies goes back to as early as 1990 [20][21][22]. A few studies also evaluated the quality of the traffic information by using NN models [23].…”

“…A few studies also evaluated the quality of the traffic information by using NN models [23]. In a more recent and relevant study, Dia and Panwai [22] analyzed the drivers' diversion given en route traffic information using both discrete choice models as well as an NN model. In this study, no attempt was made to identify the contributing factors (explanatory variables) to the diversion behavior for the undertaken case study in Australia.…”

A review of Machine Learning (ML) algorithms used for modeling travel mode choice

“…When compared to different ML algorithms that model travel mode choice, ANNs, DTs, SVMs and CA algorithms [89] FA F Other Xian-Yu (2011) [98] FA SVM BP Dia and Panwai (2010) [104] FA F BP Zhang and Xie (2008) [88] FA SVM BP Dia and Panwai (2007) [105] CLA Hybrid Other Andrade et al (2006) [87] FA F Other Celikoglu (2006) [39] CLA RBF LM Xie et al (2003) [86] CLA MLP BP Cantarella and de Luca (2003) [38] CLA MLP BP Vythoulkas and Koutsopoulos (2003) [84] FA F BP Mohammadian and Miller (2002) [83] CLA MLP BP Nijkamp et al (1996) [106] FA MLP BP Shmueli et al (1996) [81] CLA MLP BP 1 : CLA: Classification, FA: Function Approximation. 2 : RF: Random Forests, SVM: Support Vector Machines, F: Fuzzy, RBF: Radial Basis Function, MLP: Multi-Layer Perceptron.…”