Select Page

Convergence of importance sampling


This plot displays the sequence of estimates for observed log-likelihood computed by Monte Carlo approach. The purpose is to check the convergence of the algorithm.


In the following example, the parameters of a three-compartments infusion model with linear elimination are estimated on the remifentanil data set. As explained on this page, the bias of the log-likelihood estimator decreases with the number of iterations, before the estimation value stabilizes. The number of points in the plot is usually smaller than the number of iterations, and depends on the total number of observations.


  • Add/remove grid.