Table 1. Incidence prediction model diagnostics.
Singapore | Bangkok | ||||
Full | Step-down | Full | Step-down | ||
Terms | 20 | 16 | 21 | 8 | |
Model | r 2 | 0.948 | 0.948 | 0.947 | 0.943 |
Fit | Correlation | 0.931 | 0.931 | 0.879 | 0.869 |
AIC | 2760.57 | 2751.559 | 999.162 | 986.712 | |
Incidence | Lag-0 Correlation | – | 0.666 | – | 0.921 |
Prediction | Lag-4 Correlation | – | 0.785 | – | 0.762 |
Table reports the r 2, the correlation between observed dengue incidence and model fitted values, and the Akaike information criterion (AIC) for the full and step-down models in both Singapore and Bangkok. Table also reports correlation between observed 2010 dengue incidence and out-of-sample 2010 predictions from the step-down models for Singapore and Bangkok for unlagged observed incidence and observed incidence lagged by 4 weeks.