Fig. 4
From: Forecasting Chikungunya spread in the Americas via data-driven empirical approaches
Evaluation of Chikungunya transmission model failure rates, as a function of time and associated amount of data available for model calibration. Models developed with data for August 2014, September 2014, October 2014, November 2014, December 2014, and January 2015 were compared against PAHO reports in February 2015 to assess model performance from six months to one month of anticipation respectively. Models improved in terms of fit between predicted and observed Chikungunya cases where more information was included in late models. For intervention purposes, under prediction of cases (red) was more undesirable than overprediction of cases (dark blue)