Wen Chen,Xuesong Wang,Anderson Kekoa
Abstract:As the first and crucial step in improving the overall safety of a road network, hot spots identification (HSID) has been the focus of many research efforts. In general, most of the HSID methods employ either crash counts or crash rates to screen hazardous locations. However, there is a lack of empirical comparison of the two crash statistics in terms of their identification performances when dealing with HSID. To quantify the performance difference of the two crash statistics, both real road section accident data and experimentally derived simulated data are used in this paper. Three different tests are employed to evaluate different aspects of identification performance. Additionally, various levels of confidence are also explored across different situations. The results illustrate that crash counts significantly outperforms crash rates in identifying hot spots. The findings are in agreement with the other theoretical derivations shown in previous studies.