Influence of social and economic characteristics on pedestrian crash severity at mid-blocks

Vehicle-pedestrian crashes are a major concern in many urban areas. Between 2004 and 2013, about 34 pedestrians have been killed every year in traffic crashes in Melbourne Metropolitan area in Australia and vehicle-pedestrian crashes account for 24% of all fatal crashes. Although many studies have analysed the variables affecting pedestrian crashes at intersections, research on pedestrian crashes at mid-blocks is very limited. Mid-block crashes account for 46% of the total pedestrian crashes in Melbourne metropolitan area and about 50% of the pedestrian fatalities occur at mid-blocks. In addition, there are limited studies that examined the influence of different social and economic parameters in pedestrian crash analysis. This paper aims to develop a model of pedestrian crash severity to identify those influencing social and economic variables. Pedestrian postcode is used to join social and economic characteristics of suburbs to crash dataset. Boosted Decision Tree (BDT) is applied to predict the severity of vehicle-pedestrian crash at mid-blocks. In this paper, the impact of social characteristics such as level of education or type of occupation on pedestrian crash is investigated for first time. The results of this study show that the boosting technique improves the accuracy of individual Decision Tree model by 42%. The results of BDT show that social characteristics such as ethnics and level of education in suburbs are important variables influencing the severity of pedestrian crashes.