Graph feature engineering with Neo4j and Amazon SageMaker
summary
Shows how Neo4j Enterprise (v5) and Neo4j Graph Data Science produce graph embeddings with FastRP from an EDGAR filings graph (~half a million entities) and export column vectors to Amazon SageMaker; deployment uses an AWS Marketplace CloudFormation stack and a SageMaker Studio notebook, Graph Data Science is installed via pip, and SageMaker Autopilot consumes the CSV to run candidate models (Autopilot runs take ~90 minutes). Claims: graph-derived features improve binary portfolio-churn prediction versus traditional tabular features.
AI-written summary · about
chapters (6)
- 0:00Introduction to graph ML
- 0:43Architectural overview
- 2:55Deployment and setup
- 7:15Data loading and exploration
- 11:45Feature engineering demo
- 14:03Conclusion and resources
related talks
20mPower Amazon Bedrock applications with Neo4j knowledge graph
44mExploratory data analysis and automated feature engineering
30mHow Graph Data Science Can Turbocharge Your Knowledge Graph
30mAmazon SageMaker Feature Store: Store, discover, & share features for ML apps
34mPrepare data for machine learning with ease, speed, and accuracy
43mBuilding Billion Node Graphs for Machine Learning
29mAutomatically Build Models With Amazon Sagemaker
33mAccelerate data preparation with Amazon SageMaker Data Wrangler