Revolutionizing indoor emergency evacuation prediction with machine learning: A Hand-searching technique and content analysis on Random Forest algorithm

Machine learning and emergency evacuation prediction share a meaningful association in the context of the Industrial Revolution 4.0 era, given that machine learning can potentially enhance emergency evacuation prediction processes. Machine learning has the potential to revolutionise emergency evacuation prediction by providing a data-driven approach that aids decision-making and response time in emergencies. The complexity of emergency situations can lead to tragic accidents, as individual behaviour heavily influences evacuation time. While some researchers have explored emergency evacuation, existing computer evacuation models lack detailed analysis. To address this gap, this study focuses on reviewing the Random Forest algorithm, an advanced machine learning algorithm based on Decision Trees, in the application of forecasting emergency evacuation time. Through a qualitative research approach using the innovative Hand-searching technique, we conducted a systematic review employing Content Analysis Theory. Multiple sources were examined, including Google Scholar, Science Direct, Scopus, and Universiti Kebangsaan Malaysia e-Journal System. The study's findings shed light on the performance, prediction factors, advantages, and limitations of Random Forest in identifying impacts during an emergency evacuation. These insights hold significant implications for emergency responders, building designers, and policymakers.