A comprehensive literature review of the demand forecasting methods of emergency resources from the perspective of artificial intelligence

In recent decades, several forecasting methods have been proposed so as to aid in selecting from all optimal alternatives in the demand of emergency resources. Academic research in the field of emergency management has increasingly focused on artificial intelligence. However, more attention has been paid to attempts at simulating the human brain, with little focus on addressing intelligent information processing techniques based on machine learning, big data and smart devices. In this paper, a comprehensive literature review is presented in order to classify and interpret current research on demand forecasting methodologies and applications. A total of 1235 academic papers from 1980 to 2018 in the SpringerLink and Elsevier ScienceDirect databases are categorized as follows: time series analysis, case-based reasoning (CBR), mathematical models, information technology, literature reviews, and discussion and analysis. Application areas from business source premier include papers on the topics of emergency management, decision-making, decision relief, logistics, fuzzy sets and other topics. Academic publications are classified by (1) year of publication, (2) journal of publication, (3) database source, (4) methodology and (5) research discipline. The results of this literature review show that, despite forecasting methods such as ARIMA, CBR and mathematical models appearing to play a pivotal role in promoting prediction performance, there is a need to explore more real-time forecasting approaches based on intelligent information processing techniques so as to achieve appropriate dynamic demand prediction that is adaptable to emergency and rescue situations. The intention for this paper is to be a useful reference point for those with research needs in forecasting methodologies and the applications of emergency resources.