Table 1.
Summary of AICc (for training data), RMSE (for test data) and RC values for different models calculated based on the epidemic data of Shanghai. Note the negative AICc values result from the fact that data points are fewer than the free model parameters.
| Shanghai |
Early stage |
Middle stage |
Late stage |
||||||
|---|---|---|---|---|---|---|---|---|---|
| Model | AICc | RMSE | RC | AICc | RMSE | RC | AICc | RMSE | RC |
| Hill’s | 5.2 | 31 | 0.47 | 4.4 | 33 | 0.85 | 4.3 | 6.2 | 0.97 |
| Logistic | 4.4 | 120 | 0.68 | 4.2 | 10 | 0.96 | 4.3 | 4.7 | 0.99 |
| Gompertz’s | 3.9 | 25 | 0.37 | 4.2 | 34 | 0.92 | 4.6 | 6.2 | 0.98 |
| Richards’ | 4.9 | 65 | 0.73 | 4.0 | 7.8 | 0.92 | 3.7 | 2.8 | 0.99 |
| G-Logistic | 4.5 | 448 | 0.01 | 4.0 | 5.4 | 0.85 | 3.8 | 2.8 | 0.97 |
| Exp.Growth | 3.8 | 112 | 0.53 | 6.7 | 85 | 0.84 | 8.6 | 68 | 0.88 |
| Max.LLH | 4.1 | 61 | 0.21 | 7.5 | 268 | 0.11 | 9.0 | 101 | 1.0e−3 |
| Seq.Bayes. | 4.0 | 78 | 1.4e−4 | 5.1 | 13.3 | 0.39 | 6.6 | 16 | 0.71 |
| Time Dep. | 4.2 | 148 | 0.48 | 4.1 | 11.9 | 0.72 | 6.0 | 11 | 0.81 |
| SIR | 3.5 | 3.2e3 | 0.17 | 6.4 | 281 | 0.02 | 7.6 | 43 | 0.04 |
| SEIR | 3.5 | 1.1e4 | 0.76 | 6.2 | 184 | 0.11 | 7.2 | 35 | 0.60 |
| SEIR-QD | 8.9 | 5.1e3 | 1.2e−4 | 4.7 | 16 | 0.44 | 5.1 | 6.0 | 0.69 |
| SEIR-AHQ | −3.6 | 1.0e4 | 1.7e−5 | 10 | 84 | 2.8e−3 | 7.9 | 17 | 0.15 |
| SEIR-PO | −17.8 | 7.2e3 | 6.9e−5 | 5.8 | 8.2 | 0.14 | 4.9 | 3.6 | 0.83 |