A lot of operational data and decision models are based on relational data models. From a standpoint of analytics and decision making, a graph data model exposes many tools that can be leveraged for this. For example, you can deal with the violation of the Independent, Identically Distributed assumption that many of us (knowingly or unknowingly) use with ML models. The relation to graph tool is a way to transform relational data into frequently encountered models for graph based analysis.