Identification and Countermeasures of Accident Black Spot Locations Using Statistical Modeling (A Case Study of Addis Ababa)

No Thumbnail Available



Journal Title

Journal ISSN

Volume Title




Nowadays the issue of traffic safety has become the most considerable concern throughout the world, though it challenges more of developing countries, of which Ethiopia is one with its growingly urbanized capital city, Addis Ababa. With regard to traffic safety, the six-year summarized statistical data of Addis Ababa Police Commission (AAPC) revealed that 2,632 fatal, 9,133 severe injuries, 6,957 slight injuries and 85,316 property damage only (PDO) type of accidents were recorded in Addis Ababa city i.e. from July 6, 2009 – July 6, 2016. These figures pinpoint a fact how much the safety aspects of the entire road networks are becoming a pressing challenge in Addis Ababa city, and thus looking for policy attention and inquiry. The very objective of this research was therefore to investigate and develop accident prediction model using the prevalent geometric and traffic parameters. To this end, three principal arterial streets of Addis Ababa city were selected based on the premise of their susceptibility for the occurrence of frequent number of accidents and the availability of better accident data. Moreover, accident blackspot locations were distinguished through point weightage approach by employing three years of accident data from the police stations and AAPC database. Accordingly, primary data analysis was executed on the traffic and spot speed studies by employing the data gathered from site investigation survey. An accident prediction model was also derived from Generalized Linear Model (GLM), specifically from Negative binomial regression analysis using SPSS software packages, version 20. Accordingly, as per the yield of the investigations four logical factors such as traffic volume (APHV), the 85 th percentile speed, availability of U-turn and number of access points, were discovered significant, and thus incorporated in the final accident prediction model. Based on this finding, the coefficients of all variables except number of access point carry the expected signs, where they have shown positive relation with the natural log count of accident frequency. However, number of access point was found significant with negative coefficient.



Traffic Safety, Accident Frequency, Black Spot, Modeling, Geometric variables, Traffic variables, GLM, Negative Binomial