Graph Gurus Episode 29: Using Graph Algorithms for Advanced Analytics Part 3

Graph Gurus 29
Using Graph Algorithms for Advanced Analytics
Part 3 - Community Detection
1

© 2020 TigerGraph. All Rights Reserved
Today's Presenter
2
Victor Lee
Head of Product Strategy & Developer Relations
● BS in Electrical Engineering and Computer
Science from UC Berkeley, MS in Electrical
Engineering from Stanford University
● PhD in Computer Science from Kent State
University focused on graph data mining
● 20+ years in tech industry

Some Housekeeping Items
● Although your phone is muted we do want to answer your questions -
submit your questions at any time using the Q&A tab in the menu
● The webinar is being recorded and will uploaded to our website shortly
(https://www.tigergraph.com/webinars/) and the URL will be emailed
you
● If you have issues with Zoom please contact the panelists via chat
3

Move Faster with TigerGraph Cloud
4
Built for agile teams who would rather build innovative applications than
procure hardware or configure and manage databases
● Start for free
● Move to production with distributed data and HA replication

Today’s Outline
5
1
3
2
Recap of Parts 1 & 2:
Path and Centrality Graph Algorithms
Community Detection Algorithms
Clustering vs. Partitioning
Strict vs. Lenient Rules
What's a Community?
Who's in and who's out?
4
Demo
Running and modifying GSQL

Review: Analytics with Graph Algorithms
● Graph algorithms answer fundamental questions about
connected data
● Each algorithm in a library is tool in an analytics toolkit
● Building blocks for more complex business questions
6
Specialized functions Combine to make
something better

Example Questions/Analyses for Graph Algorithms
Which entity is most centrally
located?
● For delivery logistics or greatest visibility
● Closeness Centrality, Betweenness
Centrality algorithms
7
How much influence does this
entity exert over the others?
● For market penetration & buyer influence
● PageRank algorithm
Which entity has similar relationships
to this entity?
● For grouping customers, products, etc.
● Cosine Similarity, SimRank, RoleSim
algorithms
What are the natural community
groupings in the graph?
● For partitioning risk groups, workgroups,
product offerings, etc.
● Community Detection, MinCut algorithms

Summary for Shortest Path Algorithms
8
1
4
3
Graph Algorithms - tools and building
blocks for analyzing graph data
GSQL Algorithm Library - runs
in-database, high-performance,
easy to read and modify
Shortest Path Algorithms - different
algorithms for weighted and
unweighted graphs
2 Learning To Use Algorithms - know what
problem they solve, pros and cons

Summary for Centrality Algorithms
9
1
4
3
Centrality Algorithms - abstract
concepts of location and travel
Customizing GSQL Library algorithms is
easy and familiar, like procedural SQL
PageRank - uses directed referral edges
to find the most influential nodes.
Personalized PageRank is localized.
2 Closeness and Betweenness use shortest
paths. Betweenness is more complex.

Some Types of Graph Algorithms
● Search
● Path Finding & Analytics
● Centrality / Ranking
● Clustering / Community Detection
● Similarity
● Classification
10

© 2020 TigerGraph. All Rights Reserved 11
How do I find the most influential provider in each region
for a particular medical condition?
Whole-Graph Compute problem
1. Analyze claims data to identify referral relationships
among providers (Time Series Analysis)
2. Create subsets of claims around each condition with
a group of healthcare codes (e.g. CPT codes) for
each region (e.g. local healthcare market)
3. Utilize PageRank to score hubs within each market Dr. Thomas
Condition: Diabetes
Healthcare Market: S. San Jose, CA
Hub Identified: Dr. Thomas
Best Practice: Know Graph Algorithms

© 2020 TigerGraph. All Rights Reserved 12
Who is influenced by these leaders (e.g. other doctors,
chiropractors, physical therapists, facilities)?
Utilize Community Detection
1. Identify communities of providers
around each hub for each region and
for a specific condition
2. Track changes over time to detect
significant shifts in communities
Dr. Thomas
Condition: Diabetes
Healthcare Market: S. San Jose, CA
Hub Identified: Dr. Thomas
Community Detected: Diabetes – S. San Jose – Dr. Thomas

What's a Community?
What's your definition?
13
● People who live in the same
neighborhood?
● Entities which interact with one
another, in a mutual relationship
● May have things in common (similarity),
but that is not the key factor.

Community Detection
● Deciding who is in a community… and who isn't.
● If items are already labeled with their community membership,
then your work is done!
● If no rules, the we need to make
or discover our own rules,
based on level of interaction.
14

Communities can be Clusters or Partitions
● Partitioning puts each item
in exactly one group, even if
the justification is sketchy.
15
● Clustering sends up natural
boundaries. Some items
can be outside.
?

A Spectrum of Community Detection Rules
16
● Direct connection to
1+ other member of the
community
● Connected Component
K+ other members of the
community
● K-Core
every member to the
community
● Clique
K = 2

Directed vs. Undirected Connections
17
Connected Components Strongly Connected Components
A component is connected if there is a path from every vertex to every
vertex.
● If edges are directed, we say the component is strongly connected.
● If edges are undirected, we say it's (weakly) connected.

Relative Density - Louvain Modularity
18
Modularity is a measure of how "good" is the partitioning of a graph
= (fraction of the edges that fall within the given groups)
minus (the expected fraction if edges were distributed at random)
● Researchers at the University of Louvain developed an especially
efficient method for finding the partitioning with the best Q score.
Which has a higher modularity?

More Applications for Community Detection
● Marketing/Customer Analytics: find who
is chatting with whom, to understand
who is being reached and to improve
targeting.
● Biosciences: identify natural
communities of interaction at the
molecular, tissue, organism, and species
levels.
● Financial analytics: discover clusters of
transactions or contracts, to understand
and predict market dynamics, uncover
illicit activity.
19
Specialized plant biochemistry
drives gene clustering in fungi.
bioRxiv 184242; doi:
https://doi.org/10.1101/184242
http://adrianland.uk/social-media/my-l
inkedin-social-graph/
Regionalism and Overlap in Investment Treaty
Law – Towards Consolidation or Contradiction? J
of Intl Econ Law 17(2), pp. 271-298 (2014)

DEMO
GSQL Graph Algorithms in TigerGraph Cloud
20

Datasets
1. Drug Prescribers (104 vertices, 417 directed edges)
○ Demonstrate healthcare use case, visualize results
2. Kangaroos (17 vertices, 91 undirected edges)
○ to demonstrate k-core visually
○ http://konect.uni-koblenz.de/networks/moreno_kangaroo
3. Flickr Images (105,938 vertices, 2,316,948 undirected edges)
○ Demonstrate scalability of algorithms
○ http://konect.uni-koblenz.de/networks/flickrEdges
21

Data Preparation
● Graph algorithms out-of-the-box assume you have 1 type of
vertex and 1 type of edge directly connecting them:
● In our Healthcare Prescriber data set, we didn't yet have direct
Prescriber-Prescriber relationships.
● We ran a pre-processing query to induce direct Referral relationships:
22
vanilla vanilla
referral
claims
patient
prescriber date1 < date2

GSQL Graph Algorithm Library
● Written in GSQL - high-level, parallelized
● Open-source, user-extensible
● Well-documented
23
docs.tigergraph.com/graph-algorithm-library

TigerGraph GSQL Graph Algorithm Library
✓ Call each algorithm as a GSQL query
or as a RESTful endpoint
✓ Run the algorithms in-database (don't
export the data)
✓ Option to update the graph with the
algorithm results
✓ Able to modify/customize the
algorithms. Turing-complete
language.
✓ Massively parallel processing to
handle big graphs
24

Summary
25
1
3
2
Use connectedness to decide
boundaries
Strict vs. Lenient Community Rules
Black&white rules are not always helpful.
Louvain uses relative density.
Communities are Clusters, not Partitions
Don't have to include everyone.
Can overlap?
4
Pre- or Post- step with other algorithms
Many algorithms assume you start
from just one connected community

Q&A
Please submit your questions via the Q&A tab in Zoom
26

More Questions?
Join our Developer Forum
https://groups.google.com/a/opengsql.org/forum/#!forum/gsql-users
Sign up for our Developer Ofﬁce Hours (every Thursday at 11 AM PST)
https://info.tigergraph.com/officehours
27

Additional Resources
Start Free at TigerGraph Cloud Today!
https://www.tigergraph.com/cloud/
Test Drive Online Demo
https://www.tigergraph.com/demo
Download the Developer Edition
https://www.tigergraph.com/download/
Guru Scripts
https://github.com/tigergraph/ecosys/tree/master/guru_scripts
28

Upcoming Graph Guru Events
29
Coming to Atlanta, Charlotte, London, and more.
View all events and request your own here:
https://www.tigergraph.com/graphguruscomestoyou/
Supply Chain Optimisation using Native Parallel
Graphs
February 26 at 8am PST
https://info.tigergraph.com/supply-chain-optimisation

Graph Gurus Episode 29: Using Graph Algorithms for Advanced Analytics Part 3

More Related Content

What's hot (20)

Similar to Graph Gurus Episode 29: Using Graph Algorithms for Advanced Analytics Part 3 (20)

More from TigerGraph (20)

Recently uploaded (20)

Graph Gurus Episode 29: Using Graph Algorithms for Advanced Analytics Part 3