Introduction to Graph Analytics

Going from inspiration to production with graph models requires knowledge of influential and outlier nodes in the graph, clusters and communities, hidden connections between nodes, and the ability to compare different graphs based on these attributes. The Graph Analytics toolkit enables this depth of understanding by providing several methods:

• Connected components
• Graph coloring
• K-Core decomposition
• PageRank
• Single-source shortest path
• Triangle count

Each method takes an input graph and returns a model object, which contains the run time, an SFrame with the desired output for each vertex, and a new graph whose vertices contain the output as an attribute.