Clustering of time series for the analysis of the COVID-19 pandemic evolution

This study explores the dynamics of the COVID-19 pandemic by comparing the time series of ac-tive cases per population of 191 countries. Data from "Our World in Data" are examined, and Min-imal Spanning Trees and a Hierarchical Trees are used to detect groups of countries that share simi-lar performance on dynamics of coronavirus spread. Three main clusters are detected (with 104, 43 and 43 countries) and a small group composed by Mongolia and the average of all the world. The most numerous group has not reached its maximum yet and maintains a growing trend, group 2 was the first to reach the peak of daily infections and quickly entered into a phase of decline, whereas group 3 had an abrupt increase in new cases between days 20 and 40 and then entered into a de-creasing phase. The differences between the dynamics can be explained by the actions taken: there is an association between better performance and implementation of more stringent measures, as well with the realization of a greater number of tests. The results are used to discuss the dichotomy between the economic performance and health, showing that restriction policies are associated with a decrease in the number of infections. This comparative study can serve to identify the optimal public policies to minimize the number of cases and the death rate of COVID-19 in a country.