Hazard Duration Model with Panel Data for Daily Car Travel Distance: A Toyota City Case Study

Jiahang He,Takayuki Morikawa,Toshiyuki Yamamoto,Tomio Miwa

doi:10.3390/su12166331

Abstract

The limitation of battery size for electric vehicles has driven researchers to study driving distance. Trip patterns and traveler preferences in terms of distance are affected by multiple variables. This study, using socioeconomics, weather conditions, and vehicle characteristics as covariates, compares lognormal, log-logistic, and Weibull distribution assumptions on daily car travel distances with a parametric hazard model for both pooled and panel regression. The results reveal that the log-logistic distribution performed best for both the pooled and panel models, and the inclusion of heterogeneity by the panel model improves the model. The results suggest that the travel distances achieved by people in Toyota City, Japan, is highly dependent on the weather conditions, specifically the precipitation and wind speed. Socioeconomic indicators, such as age and gender, and vehicle characteristics, such as engine size and vehicle price, also significantly affect the car travel distance.

Highlights

Driving experience is obtained over time, and changes in transportation habits are typical [1]
Travel behavior has been studied by examining explanatory variables in several manners
It has been demonstrated that travel distance can be used in a survival analysis as travel time [28,32]

Summary

Introduction

Driving experience is obtained over time, and changes in transportation habits are typical [1]. Trip patterns and traveler preferences in terms of distance or time could be associated with explanatory variables such as socio-demographics and environmental characteristics. Some researchers have compared single distribution forms [22,23,24], and others have used a mixed distribution to characterize the travel distance [25,26,27] These studies of distribution functions have managed to reveal driving habits statistically but are unable to explain these habits, the specific reasons behind the driver’s behavior, but clearly some people may have the tendency of short driving distance, which makes them more adaptable for electric vehicles. The study conducted by Anastasopoulos et al [28] analyzed the distance by new energy-type transportation modes and explanatory variables, such as traveler socio-economic and demographic characteristics, using a hazard-based approach. The study of travel distance using the hazard duration approach is limited [13], and the duration dependence is often ignored since the travel distance is typically considered as a travel outcome rather than a process [29]

Methods

Results

Conclusion