Table 2. Performance for two-week-ahead forecasts of experts and non-experts.
Values have been cut to three significant digits and rounded.
| Model | Target | WIS - natural | WIS - log scale | Coverage
50% |
Coverage
90% |
||||
|---|---|---|---|---|---|---|---|---|---|
| abs. | rel. | sd | abs. | rel. | sd | ||||
| crowd-ensemble | Cases | 40.1k | 1 | 69.4k | 0.22 | 1 | 0.25 | 0.38 | 0.69 |
| Expert | Cases | 42.7k | 1.06 | 74.9k | 0.24 | 1.08 | 0.28 | 0.46 | 0.77 |
| Non-Expert | Cases | 43.1k | 1.07 | 67k | 0.26 | 1.14 | 0.25 | 0.31 | 0.54 |
| crowd-ensemble | Deaths | 40.2 | 1 | 41.5 | 0.12 | 1 | 0.07 | 0.54 | 0.77 |
| Expert | Deaths | 41.2 | 1.03 | 41.8 | 0.16 | 1.29 | 0.15 | 0.54 | 0.77 |
| Non-Expert | Deaths | 45.9 | 1.14 | 56.8 | 0.13 | 1.06 | 0.08 | 0.46 | 0.77 |