Traffic accidents represent a significant global issue, resulting in approximately 1.9 million fatalities and 50 million injuries each year, posing a grave public health threat. This problem is particularly severe in emerging countries, where street accidents rank among the leading causes of injury and death. Worldwide, governments and transportation authorities are actively seeking effective methods to reduce the severity and frequency of such incidents. To enhance our understanding of accident dynamics, this paper uses data mining techniques and statistical analysis to an extensive dataset of street accidents. It investigates several key factors influencing accident severity, including collision type, weather conditions, road surface type, lighting, and the impact of impaired driving. Machine learning methods, including the XGB Classifier, Extra Trees Classifier, Random Forest, and Decision Tree Classifier, are employed to build classification models for accident severity based on these parameters. Models are trained to classify accidents, enabling more accurate predictions of their severity. By identifying patterns within the data, these models provide crucial insights that can guide effective accident prevention strategies. The Decision Tree Classifier outperformed other models with an accuracy of 99%.