Modern ship design often involves the automated creation of thousands of design alternatives; even when provided a Pareto front of optimal solutions designers may struggle to understand the differences in designs and the relationships of design variables. Machine learning Bayesian networks from automatically developed design data can allow us to analyze the designs, understand the variable relationships that drive their differences and optimalities, and lead engineers to better designs. However, the information about variable relationships in Bayesian network are encoded in difficult to interpret conditional probability tables (CPTs). Translation of a Bayesian network’s CPTs into simpler edge weights defining the strength of relationship between nodes allows engineers to more easily interpret and use the complex information encoded in the network through standard network analysis techniques. Bayesian networks developed from a multi-objective bulk carrier design problem developed by Sen are transformed to network adjacency matrices for such analysis in this work.