Prediction of Road Traffic Accident Severity Using Machine Learning Techniques in the Case of Addis Ababa

Road traffic accidents (RTAs) are vehicle accidents that occur on a route, involving collisions between vehicles and various moving objects, such as fixed obstructions, pedestrians, other vehicles, or animals. RTAs rank as the sixth leading cause of death in Ethiopia, with approximately 31,564 lives lost to road accidents every year. The objective of this study is to analyze road accident data to predict the severity of accidents using a machine learning technique. We used 7744 records with six attributes, including the final class, to develop the model. To balance the data, the author uses SMOTE analysis. The data were split into two parts for model training and testing: 80% for training and 20% for testing. The machine learning algorithms used in this study include artificial neural network (ANN), random forest (RF), decision tree (DT), and support vector machine (SVM). The accuracy achieved by these algorithms is as follows: ANN 90%, RF 89%, DT 87%, and SVM 89%. Consequently, ANN slightly outperforms the other machine learning algorithms in predicting accident severity using traffic record data. In conclusion, machine learning algorithms demonstrate promising results in predicting the severity of accidents.