計程車駕駛者發生交通事故次數之影響因素分析

This paper investigates the factors influencing the accident likelihood and numbers of taxi drivers in Taiwan by using the data collected by MOTC in 2005. With applying the models of count data, such as Poisson, Negative Binomial and Zero-Inflated Negative Binomial model, the estimated results indicate that Zero-Inflated Negative Binomial model has better performance than others in terms of significant statistics. The factors that influence the accident likelihood and numbers of taxi drivers are "driving to find passengers", "driving mileages", "working hours per day", "aggressive driving", "maintenance cost", "insurance expense", "education level", "the years of driving taxi" and "driving in metropolitan area or not". As for the detail effects of aggressive driving behavior on the accident likelihood and numbers are appeared to be significant in "speed driving" and "illegal passing, swerving, and circumrotating".