Machine Learning and Geo-Based Multi-Criteria Decision Support Systems in Analysis of Complex Problems

Abstract

Many complex problems require a multi-criteria decision, such as the COVID-19 pandemic that affected nearly all activities in the world. In this regard, this study aims to develop a multi-criteria decision support system considering the sustainability, feasibility, and success rate of possible approaches. Therefore, two models have been developed: Geo-AHP (applying geo-based data) and BN-Geo-AHP using probabilistic techniques (Bayesian network). The ranking method of Geo-APH is generalized, and the equations are provided in a way that adding new elements and variables would be possible by experts. Then, to improve the ranking, the application of the probabilistic technique of a Bayesian network and the role of machine learning for database and weight of each parameter are explained, and the model of BN-Geo-APH has been developed. In the next step, to show the application of the developed Geo-AHP and BN-Geo-AHP models, we selected the new pandemic of COVID-19 that affected nearly all activities, and we used both models for analysis. For this purpose, we first analyzed the available data about COVID-19 and previous studies about similar virus infections, and then we ranked the main approaches and alternatives in confronting the pandemic of COVID-19. The analysis of approaches with the selected alternatives shows the first ranked approach is massive vaccination and the second ranked is massive swabs or other tests. The third is the use of medical masks and gloves, and the last ranked is the lockdown, mostly due to its major negative impact on the economy and individuals.