Connected Components

Where we left off

Every function so far started from a specific node. This lesson steps back and asks a whole-graph question instead: how many separate, disconnected pieces does this graph actually have, and is anything sitting by itself with no relationships at all?

`graph.component_stats()`: the whole-graph summary

SELECT * FROM graph.component_stats();

num_components=3  largest_component=8  num_isolated_nodes=1  total_active_nodes=11

Three separate pieces: Acme Bank's cluster (itself, its two subsidiaries, and the people who work there or lead a project), Carol's Northwind Trading (just her and her employer), and one company sitting completely alone. This is the cheapest possible check for "is my graph actually one connected thing, or several islands", useful as a sanity check after a build, before trusting that a traversal from any node can reach any other.

`graph.components()`: the pieces, ranked by size

SELECT * FROM graph.components(max_rows := 20);

Returns each component's ID and size, largest first. This course's example has three: size 8, size 2, size 1, the size-1 one is the isolated company.

`graph.isolated_nodes()`: nodes with zero edges

SELECT * FROM graph.isolated_nodes(max_rows := 100);

Specifically the nodes with *no* connections at all, not small components, genuinely disconnected. In this course's data, that's Solo Ventures (c5), a company registered as a node but never referenced by any works_at or subsidiary_of relationship. In a real dataset, a growing isolated_nodes() count over time is often a sign something upstream stopped populating a foreign key it used to.