Skip to main content

Table 3 Short-term forecasting performance in the context of the plague outbreak in Madagascar. Although the sub-epidemic model consistently outperformed the simpler models in terms of the quality of fit to the plague outbreak, the sub-epidemic model did not always outperform the Richards model based on all performance metrics in short-term forecasts. Values highlighted in italics correspond to the best performance metric at a given forecasting horizon

From: A novel sub-epidemic modeling framework for short-term forecasting epidemic waves

Model Mean absolute error (MAE) Mean squared error (MSE) Mean interval score (MIS) Percentage coverage of the 95% prediction interval
2 weeks ahead
 Sub-epidemic wave 11.3 216.8 64.4 86
 Richards 13.2 275.0 101.3 76
 Logistic 27.9 878.1 714.0 14
3 weeks ahead
 Sub-epidemic wave 12.8 353.5 90.4 86.7
 Richards 14.9 392.0 112.0 74.7
 Logistic 29.7 1003.0 792.0 12.0
4 weeks ahead
 Sub-epidemic wave 14.4 549.7 115.9 85.0
 Richards 16.4 508.0 137.9 70.0
 Logistic 31.3 1112.4 862.2 11.0
5 weeks ahead
 Sub-epidemic wave 16.3 878.5 138.8 85.6
 Richards 17.5 624.6 164.6 65.6
 Logistic 32.3 1197.0 919.1 10.4