Skip to main content
. 2021 Feb 8;4:17. doi: 10.1038/s41746-021-00384-w

Table 1.

Average mean absolute error and standard deviation (in parentheses) of forecasting models (7 and 14 days ahead) for daily deaths caused by COVID-19 in 8 countries.

Country 7 days ahead 14 days ahead
AR-F SAR-F PER-F AR-F SAR-F PER-F
US 774.49 (544.70) 545.97 (447.31) 545.06 (667.84) 1049.07 (697.66) 591.55 (454.22) 613.66 (453.73)
UK 206.08 (183.25) 128.37 (115.20) 137.89 (82.19) 370.25 (242.63) 174.77 (181.53) 227.46 (129.86)
Australia 1.48 (0.87) 0.92 (0.69) 0.97 (1.01) 1.14 (0.68) 0.94 (0.63) 1.66 (1.35)
Canada 74.61 (46.70) 55.00 (34.39) 33.89 (24.47) 108.48 (50.75) 82.36 (43.96) 52.86 (35.79)
France 354.49 (119.69) 193.81 (120.62) 151.54 (156.93) 384.38 (177.43) 245.85 (147.42) 282.03 (243.91)
Italy 207.21 (97.77) 85.77 (54.34) 85.69 (66.30) 308.99 (134.86) 162.03 (95.31) 157.86 (86.09)
Greece 1.76 (1.09) 1.43 (0.96) 1.97 (1.64) 2.14 (1.23) 1.40 (1.07) 1.86 (1.78)
South Africa 6.92 (5.85) 7.33 (6.53) 6.14 (5.58) 9.14 (7.45) 8.96 (7.25) 7.68 (6.25)
Norm. mean 0.294 (0.192) 0.198 (0.161) 0.187 (0.187) 0.382 (0.241) 0.253 (0.198) 0.266 (0.215)

The last row contains min-max normalised averages across countries, methods, and forecasting tasks to account for the different ranges in different countries. AR-F autoregressive forecasting using past deaths, SAR-F combined online search and autoregressive forecasting, PER-F persistence model.