Graph Machine Learning Tutorial

Model typed relationships with historical graph snapshots, honest link tests, sparse-node evaluation and consistent serving.

Learning path

• Graph learning contract: entities, relations and snapshot time
• Graph construction: direction, duplicates and high-degree nodes
• Neighborhood features: aggregate only the graph that existed at prediction time
• Link prediction evaluation: time splits and honest negative candidates
• Graph model evaluation: report sparse-node and relation-specific failures
• Graph serving: handle new nodes, edge deletion and snapshot swaps
• Project: evaluate a lesson-prerequisite graph recommender

Connected foundations

Use prior data and model contracts as prerequisites.

• Machine Learning Tutorial
• Recommendation Systems Tutorial

Further paths

• Anomaly Detection Tutorial
• MLOps Tutorial

Practice

• Graph Machine Learning: Projects

Continue into another subject: Knowledge Graphs Tutorial.