Overview
Explore the power of connected data in machine learning through this 44-minute webinar from Open Data Science. Learn how to leverage graph-native ML techniques to enhance predictive capabilities by incorporating relationship information. Discover different approaches to graph feature engineering, including queries, algorithms, and embeddings. Dive into ML techniques that utilize classical network science, deep learning, and graph convolutional neural networks. Gain hands-on experience in generating graph representations, creating ML models for link prediction and node classification, and applying these models to enrich existing graphs or incoming data. Understand the importance of no-code visualization and prototyping in graph-based ML. Follow along with a practical demonstration that covers exploring graphs, projecting subgraphs, calculating features, and making predictions using Neo4J and graph data science tools.
Syllabus
Introduction
Motivation
Relationships Matter
Graphs Found Form
What is Data Science
Data Science with a Graph
Graph Data Science Framework
Graph Data Science Library
Graph Algorithms
Whats New
Top New Features
Community Edition
Subgraph projections
Influence maximization
Use cases
The spectrum
Boston Scientific
Meredith Corp
AstraZeneca
Demo Overview
Exploring the Graph
Exploring the Graph in Neo4J
Projecting the Graph
Calculating Features
Fast RP Embedding
Tidy Up
Holdout
Comparison
Prediction
Predicting
Grouping
Resources
Taught by
Open Data Science