Figure 1.

(A) Logarithm of the observed $y_{\mathit{it}}$ versus the forecasted daily number of cases $y_{\mathit{it}}^{\ast }$ for each country, for up to seven days ahead, where each day ahead constitutes one panel. The forecasts were obtained from the autoregressive state-space hierarchical negative binomial model, fitted using data up to 18-November-2020. The first day ahead corresponds to 19-November-2020, and the seventh to 25-November-2020. Each dot represents a country, and the sixteen countries shown in Fig. 2 are represented by blue triangles. We add 1 to the values before taking the logarithm. (B) Observed accuracy, concordance correlation coefficient (CCC) and Pearson correlation (r) between observed ($y_{\mathit{it}}$) and forecasted ($y_{\mathit{it}}^{\ast }$) values for each of the days ahead of 18-November-2020.