Effects of Clustering COVID-19 Data Series and Its Implications

The outbreak of the highly infectious SARS-CoV-2 has been declared by the WHO as a public health emergency of international concern, since early 2020. The virus has led to a huge number of deaths and has caused major damage to the worldwide economy. In particular, countries with high reliance on the tourism sector such as the Small Island Developing States (SIDS) are continuously suffering from the COVID-19 phenomenon. Currently, there is the surfacing of vaccines to fight against the virus but the SARS-Cov 2 is passing through several mutation stages, making it more infectious and virulent. This necessitates the proper management to control the virus and in this context, Zarikas et al (2020) argued that, by clustering the countries with similar COVID-19 trend, the spread of the SARS-CoV-2 can be better managed as this can provide better guidance to some harmonised sanitary measures and policies and in monitoring the vaccination programme. In addition, this paper proposes a clustered longitudinal regression model that accommodates for some physical explanatory variables and normally distributed latent contributory effects. For this study, 18 Small Island Developing States (SIDs) were clustered in 4 groups, based on the United Nations income group classification. A clustered longitudinal autoregressive process is proposed to model the data. The Conditional Maximum likelihood approach is used to estimate the model parameters. Interestingly, different variations of the proposed process were fitted to obtain a superior fitting with lower Akaike Information Criteria.