Abstract
This paper explores motorcycle and driver characteristics associated with motorcycle accident involvement. An interview targeting motorcycle drivers was conducted in Phnom Penh. A sequential strategy including clustering and classification tree was adopted. After the groups of motorcyclist with homogeneous driving behavior are identified by cluster analysis, the classification tree was employed to establish the relationship between accident involvement and motorcycle's and driver's characteristics variables for each of the groups. To identify the importance of the variable associated with accident involvement, Random Forest was employed. The key finding here is that the job has the most significant effects on the accident involvement. Other characteristics associated with accident were also identified and discussed. Based on the discussions, several intervention measures were speculated as potential for motorcyclists' safety improvement.
