Analysis Of Traffic Accidents Using Machine Learning Under Pandemic Conditions

Authors : Ülviye Gülsüm Haşıloğlu Aras

Pages : 47-57

Doi:10.38002/tuad.1572607

View : 48 | Download : 23

Publication Date : 2025-04-30

Article Type : Research Paper

Abstract :The COVID-19 pandemic that emerged in 2019 affected all aspects of life, including spiritual, psychological, social, economic, health and transportation aspects. Despite its negative consequences, however, the COVID-19 pandemic also produced some positive results. This study investigated the effect of COVID-19 lockdowns on killed-and-injured traffic accidents in metropolitan cities and Zonguldak Province in Turkey from 2012–2019 using the Extreme Gradient Boost (XGBoost) algorithm. Nonlinear regression analyses were performed using machine learning in Python programming language on the Google Colab platform. The analysis provided an estimated number of accidents for 2020, which was compared with the real killed-and-injured accidents data from metropolitan cities and in Zonguldak in 2020. The comparison showed that COVID-19 lockdowns caused a decrease in traffic accidents in metropolitan cities and Zonguldak Province, except in Diyarbakır and Ordu. It has been revealed that the number of traffic accidents predicted by machine learning algorithms in metropolitan areas for 2020 is 18.3% higher than the number of traffic accidents in 2020. Therefore, although accurate predictions can be made with machine learning, it has been observed that there may be a margin of error in extraordinary situations such as earthquakes, wars and pandemics.
Keywords : Karantina, Trafik kazaları, Kaza Tahmin Modeli, Makine Öğrenmesi, Extreme Gradient Boost