Configure your algorithms

Recap: Running PageRank with configuration

In the previous lesson, you ran algorithms with empty configuration maps. Let’s see what it looks like to add configuration.

Here’s PageRank with its default settings (empty config):

CALL gds.pageRank.stream('actor-network', {}) // (1)
YIELD nodeId, score // (2)
RETURN gds.util.asNode(nodeId).name AS actor, score // (3)
ORDER BY score DESC // (4)
LIMIT 5 // (5)

Now let’s add configuration to change how PageRank behaves:

CALL gds.pageRank.stream('actor-network', { // (1)
  maxIterations: 40, // (2)
  dampingFactor: 0.95 // (3)
})
YIELD nodeId, score // (4)
RETURN gds.util.asNode(nodeId).name AS actor, score // (5)
ORDER BY score DESC // (6)
LIMIT 5 // (7)

Compare the two tables you returned. In both cases, Robert De Niro and Bruce Willis appear in the first two positions.

After that, some actors begin to rise up the ranks, while others fall.

Notice what changed:

The configuration map is where you control algorithm behavior. Every parameter you add customizes how the algorithm processes your graph.

Now let’s explore what configuration options are available and how to use them effectively.

Universal configuration options

Some configuration options are available across all or most GDS algorithms:

Node and relationship filtering:

Concurrency:

Execution control:

Algorithm-specific configuration

Each algorithm has unique parameters that control its behavior.

PageRank example:

Louvain example:

How configuration affects results

Let’s see how changing PageRank’s dampingFactor affects results:

Low dampingFactor (0.15):

CALL gds.pageRank.stream('actor-network', {
  dampingFactor: 0.15
}) YIELD nodeId, score
RETURN gds.util.asNode(nodeId).name AS actor, score
ORDER BY score DESC
LIMIT 20

Results emphasize direct connections. Nodes with many incoming relationships score higher, regardless of those relationships' quality.

It shouldn’t surprise us then that Jackie Chan tops this list with 153 previous credits. Robert De Niro slides down, with 123 credits to his name.

High dampingFactor (0.95):

CALL gds.pageRank.stream('actor-network', {
  dampingFactor: 0.99
}) YIELD nodeId, score
RETURN gds.util.asNode(nodeId).name AS actor, score
ORDER BY score DESC
LIMIT 20

Results emphasize relationship quality. Nodes connected to important nodes score higher, even with fewer direct connections.

In this run, Robert De Niro returns to the top, despite having fewer credits than many of those below him.

It’s fair to say, this is an accurate reflection of his importance within the movie industry.

Default dampingFactor (0.85):

Balances both factors—a good starting point for most analyses.

When to use custom configuration

Use defaults when:

Customize when:

Configuration strategy

What’s next

You now understand how to configure GDS algorithms using both universal and algorithm-specific parameters. You know when to use defaults and when to customize, and you have a strategy for tuning algorithms to your needs.

In the next lesson, you’ll practice running degree centrality in all four execution modes with different configurations.