Mastering Machine Learning with Spark 2.x

Advanced graph processing

After a quick interlude into graph generation and visualization, let's turn towards more challenging applications and more advanced techniques for graph analytics. To recap, what we have done so far in terms of graph processing is just using the basic properties of the underlying edge and vertex RDDs of a GraphX graph, as well as a few transformations, including mapVertices, mapEdges, and mapTriplets. As we have seen, these techniques are already quite useful, but by themselves not powerful enough to implement graph-parallel algorithms with. For this purpose, GraphX graph has two strong candidates, which we will discuss next. Most of the built-in GraphX algorithms, including triangle counting, PageRank and so on, ...