Title: Enhancing disaster mutual assistance decisions with machine learning: case of electricity utilities

Authors: Ali Asgary; Ghassem Tofighi; Mohammad Ali Tofighi

Addresses: Disaster & Emergency Management Program, School of Administrative Studies, Faculty of Liberal Arts & Professional Studies, York University, 4700 Keele Street, Toronto, M3J 1P3, ON, Canada ' School of Applied Computing, Faculty of Applied Science and Technology, Sheridan College Institute of Technology & Advanced Learning, Trafalgar Campus, 1430 Trafalgar Road, Oakville, Ontario L6H 2L1, Canada ' Disaster & Emergency Management Program, School of Administrative Studies, Faculty of Liberal Arts & Professional Studies, York University, 4700 Keele Street, Toronto, M3J 1P3, ON, Canada

Abstract: Disaster mutual assistance (DMA) is an important mechanism that is used by many organisations including electricity utilities to generate the needed resources during major disasters and emergencies. Decision to provide (or not to provide) mutual assistance is a complicated decision that needs to be made considering multiple factors and under time pressure and uncertainty. This paper applies several machine learning algorithms to enhance DMA decisions by electricity utilities. These methods are implemented on an experimental dataset obtained during a workshop participated by disaster management experts from several Canadian electricity utilities. Results show that all of the employed machine learning methods have very high and almost similar accuracy in predicting DMA decisions. However, Random Forest and Decision Tree provide additional information by generating the weight of each criterion, optimum thresholds that can be applied to each criterion, and visual interpretation of the decision process.

Keywords: DMA; disaster mutual assistance; machine learning; electricity utilities; random forest; decision tree; power distribution; business continuity; artificial intelligence; mutual aid; disaster response.

DOI: 10.1504/IJEM.2020.117201

International Journal of Emergency Management, 2020 Vol.16 No.4, pp.281 - 296

Received: 16 Oct 2019
Accepted: 31 Aug 2020
Published online: 23 Aug 2021 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article