Using unsupervised learning methods to group African countries based on COVID-19 prevalence

The purpose of this paper is to prepare the most commonly used cluster analysis; hierarchical and non-hierarchical cluster analysis algorithms which are the K-means, the Partition Around Medoids (PAM) and the agglomerative hierarchical, to group 37 African countries on the basis of measures of COVID-19 cases, economic development and general health resources. We found that the optimum number of clusters is four and countries like Morocco, Algeria, Libya, Gabon, Botswana, Tunisia, Mauritius, and Seychelles are grouped together into one cluster, and how Egypt and South Africa formed another cluster. A Hierarchical clustering was found to be more precise compared to the other two algorithms.