A Prediction Model of Casualties Based on Machine Learning for Selection of Fire Scenario
2021 ◽
Vol 21
(5)
◽
pp. 165-173
Keyword(s):
This study has developed a model that predicts casualties (dead and injured people) using the Classification And Regression Tree (CART). Based on the fire statistics collected over a decade, this model aims to select the appropriate risk-assessment scenarios and fire prevention and safety methods applicable on individual buildings. Our evaluation indicates that this CART model can accurately predict 48 scenarios based on 5 variables related to the types of fire, fire growth rates, and evacuation situations, and calculate the corresponding probabilities for each occurrence. This model is expected to improve future quantitative fire risk assessments.
2018 ◽
Vol 3
(3)
◽
pp. 346-355
◽
2020 ◽
Vol 42
(3)
◽
2013 ◽
Vol 48
(6)
◽
pp. 1057-1073
◽
Keyword(s):