. In the present day, people rely more heavily on transportation systems than ever before. Analysis of past accident data reveals that transportation systems consistently pose a threat to human life and property. In cases of severe accidents, it is necessary to alert an emergency center near the accident site, in addition to the police center, during the vital time period, in order to save lives and minimize casualties. Obviously, sending alert to the emergency center is not required for nonsevere accidents. This article aims to identify key features and create a stacked ensemble learning model, utilizing two models - LightGBM and XGBoost, to identify severe accidents. Based on the evaluation findings, the proposed model outperformed recent works, obtaining higher levels of accuracy (85.75%), precision (86.89%), recall (84.22%), and f1-score (85.54%).

Keywords