Graph-Powered Analytics and Machine Learning with TigerGraph

Chapter 6. Analyzing Connections for Deeper Insight

In the preceding chapters, we learned that representing data as a graph gives us the power to look more deeply and broadly across our data so we can answer questions more accurately and with more insight. We’ve looked at several use cases to see examples of how to model data as a graph and how to query it. Now we want to take a more methodical look at graph analytics. What do we mean when we say graph analytics? What are some specific techniques we can use for graph analytics?

After completing this chapter, you should be able to:

• Define graph analytics and describe what distinguishes it from general data analytics

• Understand graph analytics’ requirements and some key methods, including breadth-first search and parallel processing

• Define several categories of graph algorithms that are useful for analytics

• List a few algorithms within each category and give examples of real-world uses

Understanding Graph Analytics

Let’s start by defining data analytics in general. Data analytics is making useful observations and drawing conclusions about a body of data to help people understand the significance of the data. Analytics transforms data into useful insights. Graph analytics does the same thing, except that the structure of the data affects which data we will examine and in what order. Connections are a form of data, and the connections drive the course of the analysis.

Another distinguishing aspect of graph analytics is ...