Machine Learning Approach For Clustering Of Countries To Identify The Best Strategies To Combat Covid-19

Abstract

The purpose of this study is to identify how different government measures impacted the level of Covid-19 influence on countries of similar nature. Demographic, economic, health, and weather conditions were considered to identify countries that are inherently similar in nature. This grouping along with Covid-19 epidemiology data was used to cluster countries over a period of time after Covid-19 struck. We identified those countries which changed clusters over a time period and were influenced differently by the impact of Covid-19. We then looked at the government measures through the stringency index of containment measures and observed a relation in how different stringency measures impacted the countries differently even though they belonged to the same original group. We also observed that countries that eased restrictions quickly after containment measures had to go back to the earlier stringent measures. Gradual ease of containment measure was more efficient in tackling Covid-19. The inherent grouping of countries done in our study can be used in the future as well to deploy similar measures when faced with Covid-19 like pandemic situation. The strategies adopted on average by countries within each inherent cluster can become the base for handling Covid-19 or any such pandemic in the future. The significance of the work resides in the fact that the strategies would not be aligned to economic conditions of a nation (developed versus developing) or a single factor like healthcare facilities but based on a varied list of inherent factors using machine learning methods.