Estimating Travel Time of a Road Bottleneck Using Bus Probe Data: Toyota City, Japan

摘要

This paper utilizes low-frequency bus probe data to estimate the travel time of a road bottleneck. The probe data collected from one bus route in six months in Toyota City, Japan, are used for empirical analysis. To investigate the impact of commuting behavior of Toyota Motor Corporation (TMC) which has more than 25,800 workers in Toyota, the Gaussian mixture distribution is applied to fit four groups of bus travel time referring to the combination of working, non-working days of TMC and peak, off-peak periods on weekdays. The major findings indicate that: 1) Gaussian mixture distributions applied for peak and off-peak periods in non-working days of TMC have a bimodal feature; 2) the Gaussian mixture distribution outperforms the Gaussian distribution for the 4 categorized groups, which is indicated by a higher value of the decimal logarithm of likelihood with respect to sample data.