Human stampede causative factors and cluster risk: A multi-dimensional analysis based on ISODATA and Fuzzy Theory

Abstarct Although the interest in statistical analysis of human stampedes has increased to explore the characteristics of accidents, more effort is needed to address the complexity of causative factors and their relations to risk, due to their variability and uncertainty in each case. This study aims to examine the causative factors of human stampedes for different groups of cluster risk by introducing a combined method, the Iterative Self-organizing Clustering Algorithm (ISODATA) and Fuzzy Theory. Totally about 215 cases were collected to construct a structured dataset, and characteristics of human stampedes including accident site, accident consequence, accident type, manmade factors, hardware factors, environment factors and management factors were considered for this analysis. Results show that factors such as accident type has similarity with single-dimensional analysis, while accident site in the highest risk cluster has a different distribution with the proportion of temples 39% and schools 28% which is contrary to the highest 23% of stadiums in single-dimensional analysis. It was concluded that the site with high incidence of stampede accidents did not necessarily have a high degree of risk, due to it had a high frequency of accidents but causes little casualties. Based on model results, control points for different sites are recommended according to the characteristics of different cluster risk. This paper provides a quantitative method to compare and analyze the causative factors of different risk clusters, which has relative comparative significance and improves the accuracy of judging the risk degree that can provide basis for human stampede prevention and management.