Skip to main content
. 2021 Aug 27;12:5173. doi: 10.1038/s41467-021-25207-0

Table 1.

Detailed summary of forecast evaluation for Germany (based on ECDC data). C0.5 and C0.95 denote coverage rates of the 50% and 95% prediction intervals; AE and WIS stand for the mean absolute error and mean weighted interval score.

Germany, cases
1-week ahead incident 2-week ahead incident 1-week ahead cumulative 2-week ahead cumulative
Model AE WIS C0.5 C0.95 AE WIS C0.5 C0.95 AE WIS C0.5 C0.95 AE WIS C0.5 C0.95
epiforecasts-EpiExpert 12,333 8,781 5/10 7/10 30,329 22,157 2/9 3/9 12,334 8,781 5/10 7/10 42,667 30,669 2/9 4/9
epiforecasts-EpiNow2 11,171 7,932 5/10 7/10 37,338 27,293 4/9 7/9 11,171 7,932 5/10 7/10 47,738 34,253 4/9 6/9
FIAS_FZJ-Epi1Ger 7,798 5,709 7/10 9/10 29,190 21,058 5/9 7/9 17,255 11,264 3/10 7/10 38,925 29,937 5/9 6/9
ITWW-county_repro 34,425 28,906 0/10 2/10 64,378 53,136 0/9 2/9 34,077 28,558 0/10 2/10 101,184 84,276 0/9 2/9
LANL-GrowthRate 38,970a 23,379a 5/7 7/7 77,438a 42,294a 2/6 6/6 39,042 26,794 5/10 7/10 116,494 78,224 3/9 5/9
LeipzigIMISE-SECIR 20,019 2/5 3/5 51,115 0/4 1/4 35,901 31,690 1/10 1/10 93,111 83,670 1/9 2/9
MIT_CovidAnalytics-DELPHI 41,313a 29,004a 1/8 4/8 78,872a 61,447a 2/7 2/7
SDSC_ISG-TrendModel 10,963 10,963
UCLA-SuEIR 25,012 47,747 25,012 69,800
USC-SIkJalpha 20,028 1/1 1/1 30,891 21,567 1/1 1/1 49,640
KIT-baseline 18,475 12,998 5/10 9/10 32,690 25,543 3/9 6/9 18,475 12,998 5/10 9/10 47,472 37,155 3/9 5/9
KIT-extrapolation_baseline 12,016 10,522 7/10 10/10 36,498 26,195 6/9 7/9 12,016 10,522 7/10 10/10 47,145 35,142 7/9 7/9
KIT-time_series_baseline 15,383 11,014 5/10 9/10 44,481 28,625 4/9 8/9 15,383 11,014 5/10 9/10 61,489 39,244 4/9 8/9
KITCOVIDhub-inverse_wis_ensemble 14,017 9,358 5/10 9/10 42,063 27,993 2/9 5/9 13,464 9,265 6/10 9/10 52,972 35,194 2/9 8/9
KITCOVIDhub-mean_ensemble 16,649 10,677 4/10 8/10 42,214 27,290 1/9 6/9 15,771 10,630 4/10 8/10 57,125 37,356 1/9 6/9
KITCOVIDhub-median_ensemble 11,534 8,094 5/10 9/10 37,620 25,017 3/9 7/9 12,877 9,243 6/10 7/10 49,438 34,461 2/9 6/9
Germany, deaths
1-week ahead incident 2-week ahead incident 1-week ahead cumulative 2-week ahead cumulative
Model AE WIS C0.5 C0.95 AE WIS C0.5 C0.95 AE WIS C0.5 C0.95 AE WIS C0.5 C0.95
epiforecasts-EpiExpert 187 131 5/10 7/10 333 234 3/9 6/9 187 131 5/10 7/10 442 295 3/9 7/9
epiforecasts-EpiNow2 180 120 5/10 7/10 376 235 3/9 7/9 180 120 5/10 7/10 539 336 4/9 7/9
FIAS_FZJ-Epi1Ger 256 223 3/10 4/10 525 433 2/9 3/9 215 175 0/10 4/10 684 564 1/9 4/9
Imperial-ensemble2 254 195 5/10 5/10 252 192 5/10 5/10
ITWW-county_repro 371 355 1/10 2/10 537 483 1/9 2/9 370 354 1/10 2/10 824 757 2/9 2/9
LANL-GrowthRate 195a 128a 3/7 7/7 457a 313a 2/6 5/6 189 134 3/10 7/10 560 441 3/9 5/9
LeipzigIMISE-SECIR 621 0/5 1/5 768 1/4 1/4 1,167 991 0/10 1/10 2,184 1,799 0/9 1/9
MIT_CovidAnalytics-DELPHI 474a 357a 1/8 3/8 403a 306a 0/7 5/7 449a 331a 1/8 3/8 639a 525a 0/7 3/7
SDSC_ISG-TrendModel 357 357
UCLA-SuEIR 456 827 456 1,177
USC-SIkJalpha 489 0/1 0/1 600 500 0/1 0/1 963
KIT-baseline 479 263 2/10 9/10 835 510 0/9 5/9 479 263 2/10 9/10 1,155 707 0/9 5/9
KIT-extrapolation_baseline 202 134 7/10 9/10 383 246 5/9 8/9 202 134 7/10 9/10 493 330 5/9 8/9
KIT-time_series_baseline 238 190 6/10 9/10 624 415 4/9 8/9 238 190 6/10 9/10 866 577 4/9 8/9
KITCOVIDhub-inverse_wis_ensemble 180 114 4/10 9/10 255 147 2/9 8/9 174 108 5/10 9/10 370 224 3/9 8/9
KITCOVIDhub-mean_ensemble 204 138 3/10 9/10 298 174 2/9 8/9 216 147 3/10 9/10 441 262 2/9 8/9
KITCOVIDhub-median_ensemble 200 135 4/10 8/10 334 216 3/9 7/9 202 133 4/10 8/10 440 270 3/9 8/9

aEntries where scores were imputed for at least 1 week. Weighted interval scores and absolute errors were imputed with the worst (largest) score achieved by any other forecast for the respective target and week. Models marked thus received a pessimistic assessment of their performance. If a model covered less than two-thirds of the evaluation period, results are omitted.