Analyzing heterogeneous accident data from the perspective of accident occurrence

The likelihood of achieving quantified road safety targets: a binary logistic regression model for possible factors

In past several decades, many countries have set quantified road safety targets to motivate transport authorities to develop systematic road safety strategies and measures and facilitate the achievement of continuous road safety improvement. Studies have been conducted to evaluate the association between the setting of quantified road safety targets and road fatality reduction, in both the short and long run, by comparing road fatalities before and after the implementation of a quantified road safety target. However, not much work has been done to evaluate whether the quantified road safety targets are actually achieved. In this study, we used a binary logistic regression model to examine the factors - including vehicle ownership, fatality rate, and national income, in addition to level of ambition and duration of target - that contribute to a target's success. We analyzed 55 quantified road safety targets set by 29 countries from 1981 to 2009, and the results indicate that targets that are in progress and with lower level of ambitions had a higher likelihood of eventually being achieved. Moreover, possible interaction effects on the association between level of ambition and the likelihood of success are also revealed.

Factor complexity of crash occurrence: An empirical demonstration using boosted regression trees

Factor complexity is a characteristic of traffic crashes. This paper proposes a novel method, namely boosted regression trees (BRT), to investigate the complex and nonlinear relationships in high-variance traffic crash data. The Taiwanese 2004-2005 single-vehicle motorcycle crash data are used to demonstrate the utility of BRT. Traditional logistic regression and classification and regression tree (CART) models are also used to compare their estimation results and external validities. Both the in-sample cross-validation and out-of-sample validation results show that an increase in tree complexity provides improved, although declining, classification performance, indicating a limited factor complexity of single-vehicle motorcycle crashes. The effects of crucial variables including geographical, time, and sociodemographic factors explain some fatal crashes. Relatively unique fatal crashes are better approximated by interactive terms, especially combinations of behavioral factors. BRT models generally provide improved transferability than conventional logistic regression and CART models. This study also discusses the implications of the results for devising safety policies.

Spatial-temporal analysis of drink-driving patterns in Hong Kong

Normally, bars and restaurants are the preferred locations for drinking. Therefore, there is concern that the roads in bar and restaurant areas could have a higher probability of drink-drivers and alcohol-related road crashes. Many studies have been conducted to model the association between drinking locations and the prevalence of drink-driving, so that cost-effective enforcement strategies can be developed to combat drink-driving. In this study, a cluster analysis approach was applied to model the spatial-temporal variation of drink-driving distribution in Hong Kong. Six spatial-temporal clusters of drink-driving distribution emerged from the data: (i) bar and restaurant area, weekend-overnight; (ii) bar and restaurant area, other timespan; (iii) urban area, weekend-overnight; (iv) urban area, other timespans; (v) rural area, weekend-overnight; and (vi) rural area, other timespans. Next, separate zero-inflated regression models were established to identify the factors contributing to the prevalence of drink-driving for each of the six recognized clusters. The results indicated that drivers in rural areas tend to consume more alcohol than those in urban areas, regardless of the time period. In addition, both seasonal variation and vehicle class were found to determine the breath alcohol concentration (BrAC) levels among drivers.

The impact of traffic violations on the estimated cost of traffic accidents with victims

We analyse accidents with victims and calculate the influence of traffic violations on the probability of having a serious or fatal accident, compared to a slight accident. Traffic violations related to speed limitations, administrative infringements or faults related to the driver are considered. Data were obtained from all available reports on accidents with victims that occurred in Spain from 2003 to 2005. A multinomial logistic regression model is specified to find the probability that an accident with victims is slight, serious or fatal, given the presence/absence of thirty different types of traffic violations. The average cost per victim and the average number of victims per accident are then used to find the estimated cost of an accident with victims, given the information on the traffic violations incurred. This demonstrates which combinations of traffic violations lead to higher estimated average costs, compared to cases in which no traffic violation occurred. We conclude with some recommendations on the severity of penalties, and suggest that regulators penalize the occurrences of some specific combinations of traffic violations more rigorously.