Figure 4.
Renewal model selection. We simulate 
epidemics from renewal models with 
and 
. We test the ability of several model selection criteria to recover the true 
from among this set. Each epidemic has an independent, piecewise-constant 
, examples of which are shown in (a). These models change in amplitude but not 
for every simulation. Panel b) shows the probability of detecting the true model as a function of 
and (i) considers 
with 
while (ii) uses 
and 
. The FIA performs best at every 
in (i) and overall in (ii).