Predictive Model Assessment for On-time Emergency Response and Decision Making
Open AccessThe effectiveness of emergency management operations is measured by their ability to respond to emergency incidents in a manner that prevents the loss of life. The National Fire Protection Agency (NFPA) maintains guidelines for critical time threshold that are industry standards used by emergency responders. For this study, the threshold is set to 10 minutes, which is the time between when the dispatcher notifies the responder about the incident and when the responder arrives at the scene (call preparation and arrival time). This study involves the use of various machine-learning regressors, analyzing and comparing them to predict emergency response within 10 minutes; the trees methods having the worst performance of all the models. The boosting methods have better performance at predicting response time; the histogram gradient boosting regressor is the model with the best performance. The use of a loss function with the histogram gradient boosting method ensures the generalization and optimal performance of the model on new data. This study also involves the creation of a decision support tool using the significant variables and machine-learning classifiers. The histogram gradient boosting classifier has the best performance out of the classification models. The aim of this praxis is to help emergency operators make better decisions at the call preparation (also called call processing) stage by outlining the variables that are important in managing response time. Areas for further research include studying the use of stacked methods for prediction to increase prediction performance.
- All rights reserved
Notice to Authors
If you are the author of this work and you have any questions about the information on this page, please use the Contact form to get in touch with us.