Graphical model estimation performance of the proposed method (BVS-SL) approach under different values of the belief parameter, in the case where the mixed covariates consist of continuous and binary variables, where the binary predictors are generated as in Eq (5) under logit and probit links.

The results for different precision matrices structures as in Cases I(a)-(d), are presented in terms of specificity and sensitivity under the estimated graph, and the error in estimating the precision matrix in terms of the Frobenius norm.