Figure 1.

Illustration 1—Mean square errors (MSE) for estimating correlation parameters based on partial samples for $N_{Total}\in \{50,75,100,200\}$ (number of variables in the covariance matrix) and $N_{kernel}\in \{10,20,30,100\}$ (number of incomplete covariance matrix samples). Each incomplete covariance matrix had a random size between 10 to 40. The MSEs are calculated over 10 replications of the experiment.