Towards safer educational Facilities: Computational and observational analysis of evacuation processes in primary schools

Abstract

Unlike adult or mixed-age studies, this research examines how young children's limited spatial awareness, evolving risk perception, and reliance on adults influence their evacuation choices and timing under varied signage, guidance, layouts, and visibility. Combining RReliefF feature selection with two machine learning frameworks and inferential tests, we used a unique CCTV-based dataset of 742 six-to-seven-year-olds from ten real-world drills to determine the most important predictors. The study reveals that school corridors are the most critical place affecting the children's evacuation process. Teacher presence in both classrooms and corridors significantly accelerates response times, while seating location affects students' exit choices from the classroom, and visibility is crucial in corridor evacuation time and exit selection. By providing practical suggestions, customised training, design modifications, and staff drills to enhance school evacuation safety, the study fills a vital gap in emergency-response literature and lays the groundwork for future research to refine these strategies further.