2018 Volume 4 Issue 1 Pages A_196-A_205
The objective of the paper is to construct rest area choice model and rest time model during a trip on inter-urban expressways by using ETC 2.0 probe data and attribute data of rest areas. First, principal component analysis is conducted with regard to the attributes of rest area size and shop types that are assumed to affect driver's resting behavior during their trips, and as a result, three components are selected to feature rest areas. Secondly, as a binomial choice problem of pass or drop-by when a driver comes to a rest area, the paper constructs several disaggregate logit models with independent variables of trip contextual information and three components of rest area attribute characteristic, and also estimates model parameters with high likelihood ratio. It is found with the identified models that time period dummy variable tends to distort the model structure and consequently reduces the model accuracy. Finally, a rest time model is formulated as a composite Gamma distribution of short rest time and long rest time in consideration of the existence of dining facilities and time period that may affect the rest time.
