Pipeline catalog

The Neo4j Graph Data Science library offers the feature of machine learning pipelines to design an end-to-end workflow, from graph feature extraction to model training. The pipeline catalog is a concept within the GDS library that allows managing multiple training pipelines by name.

Once created, a pipeline is stored in the pipeline catalog. When configuring a pipeline, it is resolved from the catalog and modified with the requested configuration, such as adding a training method. A pipeline is used to train a machine learning model which is stored in the Model catalog.

The different kinds of pipelines supported by GDS are described elsewhere in this chapter. This section explains the available pipeline catalog operations:

Name Description

gds.pipeline.list

Prints information about pipelines that are currently available in the catalog.

gds.pipeline.exists

Checks if a named pipeline is available in the catalog.

gds.pipeline.drop

Drops a named pipeline from the catalog.