Fig 1.
Changes in the first and second eigenvalues of COVID-19 deaths and confirmed cases.
The largest variances in the data can be observed at the beginning of COVID-19 during April and May.
Fig 2.
Components of eigenvectors (a) the first eigenvector for death cases, (b) second eigenvector for death cases, (c) the first eigenvector for confirmed cases, and (d) the second eigenvector for confirmed cases. The horizontal axis shows the month of the last data point of each time window. The vertical axis represents the affected countries. The maroon shaded areas indicate dominating eigenvectors of different affected countries.
Fig 3.
Scatter plots of PC1 and PC2 coefficients of the top 25 countries affected by COVID-19 (a) during April 2020 for confirmed cases, (b) April 2020 for death, (c) May 2020 for confirmed cases, (d) May 2020 for death, (e) October 2020 for confirmed cases, (f) October 2020 for death, (g) February for confirmed cases, and (h) February for deaths. We showed only the periods where group dynamic show significant meaning. Countries make strong cluster in the loading plots when the states of the diseases are similar and severe among the countries.
Fig 4.
Time varying change of PC for (a) COVID-19 death cases and (b) confirmed cases. The change of color indicates the change of the states of the diseases.