Predicting Motorbike Accident Severity in Nigeria Using XGBoost Machine Learning Algorithm
Authors: Olayinka Olusegun Lawal, Michael Segun Olajide, Abraham Temilade Olumide and Oluwafemi Omoniyi Samuel AbeNigeria is a developing country in which Motorbike accidents are common, as one of the causes of injury and mortality. Predicting the severity of these accidents using machine learning is essential for managing their incidence and effectively reducing casualty figures. The dataset on motorbike accidents from 2018 to 2023 was obtained from the Nigerian Federal Road Safety Corps (FRSC). The study adopted multiple classification algorithms, notably Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM), and Extreme Gradient Boost (XGBoost), to develop and implement models that predict and assess the severity of motorbike accidents. The dataset was preprocessed and split into 80% for training and 20% for testing. The accuracy, precision, recall, F1-score, and AUC metrics showed that the XGBoost model outperformed the baseline models, achieving 96.7% accuracy, an F1-score of 0.95, and an AUC of 0.97. The results showed that the Extreme Gradient Boost (XGBoost) machine learning technique provided valuable insights to help FRSC, traffic officers (Nigeria Police Force), traffic managers, and policymakers reduce the severity of motorbike accidents in Nigeria.

