Hot-Spot Identification: a Categorical Binary Model Approach

Abstract

An alternative methodology is presented for hot-spot identification based on a probabilistic model. In this method, the ranking criterion for hot-spot identification conveys the probability of a site's being a hot spot or not being a hot spot. A binary choice model is used to link the outcome to a set of factors that characterize the risk of the sites under analysis on the basis of two categories (0/1) for the dependent variable. The proposed methodology consists of two main steps. After a threshold value for the number of accidents is set to distinguish hot spots from safe sites (Category 1 or 0, respectively), a binary model based on this classification is applied. This model allows the construction of a site list ordered by using the probability of a site's being a hot spot. In the second step, the selection strategy can target a fixed number of sites with the greatest probability or all sites exceeding a specific probability, such as .5. To demonstrate the proposed methodology, simulated urban intersection data from Porto, Portugal, covering 5 years are used. The results of the binary model show a good fit. To evaluate and compare the probabilistic method with other commonly used methods, the performance of each method is tested by its power to detect true hot spots. The test results indicate the superiority of the proposed method. This method is simple to apply, and critical issues such as assumptions of a prior distribution effect and the regression-to-the-mean phenomenon are overcome. Further, the model provides a realistic and intuitive perspective.