Advanced Analytics¶
There are quite a few advanced uses of TigerGraph. Let's examine a few of them!
More exciting topics to come soon 😄😄
Graph + Machine Learning¶
Originally sourced from TigerGraph's Ecosys Docs, produced by Johnathan Herke
You might be thinking why Graphs for Machine Learning and AI? There are a few reasons including that it's a natural data model, we think in terms of graphs which means graphs can be modeled very logically and naturally.
Graphs also provide richer data by having connections between entities which allow us to create graph-based features. Graphs have always had a natural role in machine learning if you look at unsupervised learning through graph algorithms, frequent pattern mining and learning through neural networks, deep learning you will see the correlations with graph.
Graph data models are uniquely qualified to provide explanatory AI which through traditional ML techniques is hard to provide. Native graphs with Massively Parallel Processing - (like TigerGraph) enable large scale feature extraction and in-graph analytics which can be more challenging with many joins in a relational database.
Let's Explore Five Categories¶
1. Unsupervised Learning with Graph Algorithms
2. Feature Set Extraction for Machine Learning
3. ML Enrichment with Graph Features
4. Graph Enrichment with Machine Learning
5. In-database ML Techniques for Graphs
Unsupervised Learning with Graph Algorithms¶
Overview¶
Using Graph Algorithms for Advanced Analytic
Hands On Project¶
In this exercise, we will load a small graph with natural communities into the TigerGraph graph platform, install the Louvain Modularity algorithm from the graph algorithm library, run the algorithm, and visualize the result.
Feature Set Extraction for Machine Learning¶
Overview¶
Accelerating Data Science with Python + TigerGraph
Graph Features Extraction in Business Use-Case
Hands On Project¶
Machine Learning TigerLab
tensorflow google colab python pandas
Hands On Project 2¶
In this exercise, we take a synthetic dataset for retail purchase fraud, model it as a graph, extract various graph-based and non-graph-based features, export the feature vectors, and use conventional ML techniques to train a fraud prediction model. The graph model enables the
This exercise is taken directly from an exercise in Andrew Ng's Coursera online course on Machine Learning.
Machine Learning Enrichment with Graph Features¶
Overview¶
Integrating Real-Time Deep-Link Graph Analytics with Spark AI
Graph Enrichment with Machine Learning¶
Overview¶
Combining Natural Language Processing with a Graph Database for COVID-19 Dataset
Hands on Project¶
If you would like to follow along and conduct this workshop outlined above, here are the resources to get started. Please reference the video above as an instruction set for the project.
In Database Machine Learning¶
Overview¶
In-Database Machine Learning Solution For Real-Time Recommendations
Deep Learning Implemented by GSQL on a Native Parallel Graph Database
Hands on Project¶
Big Data Entity Resolution Starter Kit Demo