Assessment of COVID-19 Pandemic using Multivariate Data Analysis: A Case for Nigeria

Background: The reduction of dataset dimensions for better presentation, visualization, postulation testing, and clarification has not been reported by researchers in Nigeria COVID-19 cases. To realize the impact and magnitude of coronavirus (COVID-19) pandemic, univariate statistical analysis is monotonous in describing daily reported datasets. However, to truly understand the events and contrasting outcomes for different states in developing nations like Nigeria, multivariate analytical tools were applied to envisage and differentiate states more specifically and precisely. We made use of data analysis tools that can display the arrangement of data points in fewer proportions while keeping the dissimilarity of the original data to a feasible extent, and cluster states according to their results on the established proportions.

Methods: Pearson's correlation coefficient, principal component analysis, and hierarchical cluster analysis analyzed the data from 36 states of Nigeria and the FCT-Abuja; with COVID-19 cases for 6 months, eligible states included in the analysis are those with total cases of 20 or more with no irregular data.

Results: After performing Pearson's correlation coefficient, it reveals that the month of July, August, and October correlated with the discharged cases and active cases among the states but increased in the last quarter of 2020. The principal component analysis identified that total death emerged as the principal component for Lagos state was from October through December while Delta, Edo, Rivers, and Kano state were behind. Hierarchical cluster analysis associated the EndSars protest to have equally contributed to the striking clusters in some other states like Rivers, Enugu and Delta.

Conclusions: In summary, accurate multivariate analyses have shown to be of great value and can simplify concepts, impacts, communicate research findings, and reinforce decision-taking and public-policy drivers.