This is the companion repository for my Data Science & Analytics MBA Term Paper: Social Network Analysis applied to Data Warehouses: opportunities and constraints for Data Governance.
The image below illustrates a significant result of the research, which is the acknowledgment that the tables and connections in the assessed Data Warehouses have characteristics of Social Networks.
In a nutshell, the discreteness of the entities and the binarity of the relationships are what matters for network analysis (Zinoviev, 2018). Each circle in the image represents a table as a discrete entity that is separable from all other tables in a given data warehouse. Each arrow represents a relationship involving two discrete entities. The graphical representation leverages the nodes' out-degree attribute to highlight outstanding nodes and express how different the nodes are according to this metric.
Zinoviev (2018) also states that a complex network has a non-trivial structure. It is not a grid, not a tree, not a ring—but it is not entirely random, either. Complex networks emerge in nature and the man-made world as a result of decentralized process with no global control. One of the most common mechanisms is the preferential attachment (Emergence of Scaling in Random Networks), whereby nodes with more edges get even more edges, forming gigantic hubs in the core, surrounded by the poorly connected periphery. Such a behavior can be observed in the above image, not only by the size of the nodes but also by the density of overlapping edges in some areas of the chart.