Accident Hotspot Detection

Road accidents endanger public safety and need comprehensive procedures to detect and minimize accident-prone regions. To decrease the possibility of accidents occurring, numerous areas must be relooked at and replanned in such a manner that the likelihood of collisions is reduced, and more lives are saved. Road traffic injuries claim at least one life every 24 s worldwide, making them the eighth biggest cause of mortality. The research paper presents a novel approach visualizing an area’s accident hotspots using the clustering approach mini-batch K-means algorithm, which will assist the government in gaining a better understanding of accident-prone regions. By leveraging a large-scale dataset containing historical accident data, we aimed to efficiently identify regions with a higher likelihood of accidents, allowing for targeted interventions to enhance road safety. The proposed algorithm achieved the silhouette score of 0.33.