Highway accident number estimation in Turkey with Jaya algorithm

Abstract

In the transportation sector in Turkey, approximately 90% of cargo and passenger transportation is carried out on highways. In recent years, increasing population and welfare levels have brought along an increase in demand for and intensity of highway use. Accidents experienced along with the increased intensity in the use of highways result in fatalities and loss of property. In order to minimize such losses on the highways and determine plans and programs for the future by benefiting from historical data, it is necessary to conduct accurate, consistent, effective, and reliable accident estimations. In the study, highway accident number estimation (HANE) in Turkey was made by using the meta-heuristic Jaya optimization algorithm. For HANE, Jaya linear (Jaya-L) and Jaya Quadratic (Jaya-Q) models were proposed. Indicators such as the number of accidents that occurred between 2002 and 2018, population, gross domestic product (GDP), total divided road length (TDRL), and the number of vehicles were taken for HANE. Indicators were analyzed for four different conditions. HANE was made by using Population–GDP–TDRL–Number of Vehicle indicators together. A total of 75% of the total 17-year data between 2002 and 2018 were used for training purposes, and 25% of the data were used for testing. The results of the proposed Jaya-L and Jaya-Q models were analyzed by comparing them with the Andreassen estimation model (AEM) and multiple linear regression (MLR) methods. Following the successful training and testing results, low, expected, and high scenarios were proposed, and the number of accidents between 2019 and 2030 was estimated. © 2022, The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature.