Table 1. Mean absolute errors from cross-validation across three seasons for the “nowcasting” task.
Model | 11-12 | 12-13 | 13-14 | |
---|---|---|---|---|
(a) | Revised CDC (y) | 0.10 | 0.24 | 0.24 |
(b) | Original CDC (~y) | 0.20 | 0.30 | 0.32 |
(c) | Twitter (z) | 0.33 | 0.36 | 0.48 |
Twitter (z) + Original CDC (~y) | 0.14 | 0.21 | 0.21 | |
(d) | GFT (z) | 0.35 | 0.71 | 0.89 |
GFT (z) + Original CDC (~y) | 0.20 | 0.45 | 0.28 | |
(e) | Twitter + GFT (z) | 0.24 | 0.67 | 0.62 |
Twitter + GFT (z) + Original CDC (~y) | 0.15 | 0.33 | 0.21 | |
(f) | Historical Average (~y) | 0.95 | 0.87 | 1.39 |