Jordan Journal of Civil Engineering

Modeling Risk of Road Crashes Using Aggregated Data


Randa Oqab Mujalli;


Traffic crashes constitute a major burden on most governments, especially in countries classified as middle- or low-income countries. Most low- or middle-income countries suffer either from scarce or no crash records’ data or unreliable data, where annual crash reports become the main source of data available to investigate crashes in order to find and take countermeasures to reduce both frequency and severity of crashes. This paper aimed at using aggregate annual crash reports for 18 years, in order to determine the main factors that contribute to characterizing crashes in a specific year according to severity. Identification of these factors was made possible using Bayesian networks, in which three different models were developed. The main contributing factors which were found to increase the likelihood of classifying crashes as severe or fatal were: higher number of traffic control device violations, speeding, higher number of run-off-road crashes and higher number of pedestrian crashes.


Crash annual reports, Classification, Severity, Aggregated data