Figure 1
From: Automated parameter estimation for biological models using Bayesian statistical model checking
Comparison of the efficiency of (a) Bayesian and (b) SPRT hypothesis testing. In both cases, the number of samples required for hypothesis testing increases as the specification threshold probability approaches the actual probability with which the model satisfies the specification. Bayesian hypothesis testing required fewer samples than the SPRT when the model is obviously flawed with respect to the desired behavior. The number of samples for the Bayesian hypothesis testing vary from 1 to 1000 while those for the SPRT become as large as 100000.