Abstract: The leading cause of death that halts socioeconomic advancement in society is a traffic accident. Nigeria is one of the nations that has experienced a rise in road accidents as a result of a number of contributing factors. In order to predict the severity of road crashes in Sokoto and identify the factors that produce accurate predictions, a comparative analysis will be done using four machine learning techniques, including Decision Tree (DT), Support Vector Machine (SVM), K Nearest Neighbor (KNN), and Nave Bayes (NB). This study will employ data from the Federal Road Safety Corps (FRSCN), Sokoto command. Using the Waikato Environment for Knowledge Analysis (WEKA), the experiment will be carried out. The final result for the experiments shows that Random forest (RF) has the highest accuracy score with 98.11% followed by Support Vector Machine (SVM) with the accuracy score of 94.33%, followed by K Nearest Neighbour (KNN) with the accuracy of 92.45% and the last model with the lowest score is Naïve Bayes with accuracy score of 84.90%.

Keywords: Accident Severity, Machine Learning, Naïve Bayes, Weka, Data mining.

| DOI: 10.17148/IARJSET.2024.11901