Graph Data Science in Practice
Hands-on graph data science workshop: apply Louvain, WCC, and degree centrality graph algorithms to detect fraud communities in a Neo4j sandbox.
In this 4-hour workshop, you will learn
Go to glossary for graph data science (opens in a new tab)Analysing data through the structure of its connections. Also the name of the Neo4j library that implements it. in Practice is the companion course to an instructor-led GraphAcademy workshop. Over roughly four hours, you learn the foundations of the Neo4j Graph Data Science (GDS) library and then apply graph algorithms to a real business problem: finding fraud in a financial network.
The workshop is built around one idea: graph algorithms are only useful when they answer a business question. The first half gives you the mechanics of graph data science — Full definition for projection (opens in a new tab)An in-memory copy of part of your database that graph algorithms run against. You choose which nodes and relationships it holds., algorithm categories, and execution modes. The second half puts them to work, using Full definition for community detection (opens in a new tab)A family of algorithms that group nodes by how they connect. Each algorithm defines a community differently. algorithms such as Full definition for Louvain algorithm (opens in a new tab)A community detection algorithm that repeatedly merges nodes into groups for as long as merging raises modularity. and Full definition for weakly connected components (opens in a new tab)Groups of nodes in which every node is reachable from every other, once relationship direction is ignored. The GDS algorithm that finds them takes the same name. (WCC) to surface clusters of suspicious accounts that no single query could find.
Graph Data Science
Apply the Neo4j Graph Data Science library to real-world data, running graph algorithms in a sandbox environment to extract meaningful analytical insights.
Community Detection
Use Weakly Connected Components and the Louvain algorithm to identify clusters and communities within a graph, understanding how nodes group by relationship patterns.
Fraud Detection
Build a fraud detection pipeline by combining degree centrality and community detection algorithms to surface suspicious accounts and transaction clusters.
What you need to take part.
A GraphAcademy account
The workshop is delivered here on GraphAcademy, so you need to be signed in to work through the lessons and keep your progress. Creating an account is free.
2 modules, 4 hours.
Who this workshop is for
This course accompanies a live, instructor-led workshop. If you are attending one, this is your workbook: the lessons follow the session, and you run every query alongside the instructor. You can also work through it on your own at your own pace. It suits analysts, data scientists, and developers who know their way around Neo4j and want a practical, problem-first introduction to graph analytics — you need no prior experience with the Go to glossary for graph data science (opens in a new tab)Analysing data through the structure of its connections. Also the name of the Neo4j library that implements it. library, but you should complete Neo4j Fundamentals and Cypher Fundamentals first, since Full definition for projection (opens in a new tab)An in-memory copy of part of your database that graph algorithms run against. You choose which nodes and relationships it holds. and algorithms build directly on graph concepts and basic Full definition for Cypher (opens in a new tab)Neo4j's implementation of GQL, the ISO standard query language for graph databases. It is declarative: you describe the pattern to find, and the database decides how to find it.. Fraud detection is the worked example, but the same Full definition for community detection (opens in a new tab)A family of algorithms that group nodes by how they connect. Each algorithm defines a community differently. approach applies wherever the interesting signal lives in how records connect, from money laundering rings to bot networks.
What you'll do
This is a hands-on workshop. In the first lesson you create a Neo4j Sandbox from the Fraud movies template, a cloud-hosted instance pre-loaded with the Go to glossary for graph data science (opens in a new tab)Analysing data through the structure of its connections. Also the name of the Neo4j library that implements it. library and the movies and fraud datasets used throughout the session. Every query in the workshop runs directly in that sandbox with no further setup.
You will project graphs, run algorithms in different execution modes, and configure Full definition for projection (opens in a new tab)An in-memory copy of part of your database that graph algorithms run against. You choose which nodes and relationships it holds. to match each algorithm's requirements. In the second half you take on the fraud problem end to end: reducing a large Full definition for transaction (opens in a new tab)A unit of work that either succeeds in full or leaves no trace. network to a manageable search space with Full definition for Louvain algorithm (opens in a new tab)A community detection algorithm that repeatedly merges nodes into groups for as long as merging raises modularity., adding Full definition for degree centrality (opens in a new tab)A score for each node equal to its number of outgoing relationships. and Full definition for weakly connected components (opens in a new tab)Groups of nodes in which every node is reachable from every other, once relationship direction is ignored. The GDS algorithm that finds them takes the same name. to the investigation, and assembling communities of likely fraudsters. By the close of the workshop you will have carried a graph data science analysis from raw connected data to a ranked list of suspicious communities — the shape of most real graph analytics projects.
Where to go next
The direct sequel is Analyze Graph Data with Python, the follow-on workshop that takes the same Go to glossary for graph data science (opens in a new tab)Analysing data through the structure of its connections. Also the name of the Neo4j library that implements it. concepts out of Neo4j Browser and into the Python client, then scales up with Go to glossary for Aura Graph Analytics (opens in a new tab)The Aura service that runs graph algorithms in a separate session, with no plugin to install.. For a deeper, self-paced treatment of the foundations covered here, take Get started with Graph Data Science. When you are ready to certify your skills, the Neo4j Graph Data Science certification validates exactly the loop you practised on the fraud problem. Upcoming instructor-led sessions are listed on the workshops page.