Skip to main content
. Author manuscript; available in PMC: 2013 Jan 3.
Published in final edited form as: Biometrics. 2012 Apr 16;68(4):1238–1249. doi: 10.1111/j.1541-0420.2012.01757.x

Figure 2.

Figure 2

Comparison of the computational efficiency between the traditional MCEM algorithm and the hybrid EM-MCEM algorithm for disease transmission in independent households. Panel (a) shows the distribution of δh, the numbers of possible realizations of Uh, across all households over 100 simulated epidemics. Black diamonds along the horizontal axis indicate the values of J used in other panels, Panels (b), (c) and (d) plot the average computational times of importance sampling (black), EM iterations (white) and variance calculation over simulated epidemics versus J, the minimum value of δh to apply the MCEM algorithm.