Hypergraph Visualization Examples¶
This guide provides comprehensive examples of visualizing hypergraphs using Hypergraph-DB's built-in visualization capabilities.
β‘ Important: Code Execution Order¶
When using visualization features, please pay attention to the following code execution order to ensure you see complete analysis results:
# β
Recommended code organization
# 1. Create data
hg = HypergraphDB()
hg.add_v(...)
hg.add_e(...)
# 2. Perform analysis first (before visualization)
print("Analysis results:")
print(f"Network size: {hg.num_v} vertices, {hg.num_e} edges")
# Other analysis...
# 3. Start visualization last (program ends when user presses Ctrl+C)
print("Starting visualization...")
hg.draw() # Blocks until user presses Ctrl+C
# β Code organization to avoid
hg.draw() # Program exits immediately when user presses Ctrl+C
print("These analysis results will never be shown") # Never executed
π― Example 1: Social Network Analysis¶
Let's create and visualize a social network where groups of friends participate in various activities together.
from hyperdb import HypergraphDB
# Create the social network hypergraph
social_network = HypergraphDB()
# Add people as vertices
people = {
"alice": {"name": "Alice", "age": 25, "city": "New York", "interests": ["reading", "music"]},
"bob": {"name": "Bob", "age": 27, "city": "San Francisco", "interests": ["sports", "travel"]},
"charlie": {"name": "Charlie", "age": 23, "city": "Boston", "interests": ["coding", "gaming"]},
"diana": {"name": "Diana", "age": 26, "city": "Seattle", "interests": ["art", "photography"]},
"eve": {"name": "Eve", "age": 24, "city": "Austin", "interests": ["music", "cooking"]},
"frank": {"name": "Frank", "age": 28, "city": "Denver", "interests": ["hiking", "travel"]}
}
for person_id, info in people.items():
social_network.add_v(person_id, info)
# Add social activities as hyperedges (connecting groups of friends)
activities = [
# Small gatherings
(("alice", "bob"), {
"activity": "Coffee meetup",
"date": "2024-01-15",
"location": "Central Park",
"duration": 2
}),
# Medium group activities
(("alice", "charlie", "eve"), {
"activity": "Music concert",
"date": "2024-01-20",
"location": "Madison Square Garden",
"duration": 4
}),
# Large group activities
(("bob", "diana", "frank", "eve"), {
"activity": "Hiking trip",
"date": "2024-02-01",
"location": "Yosemite National Park",
"duration": 48
}),
# Full group gathering
(("alice", "bob", "charlie", "diana", "eve", "frank"), {
"activity": "Birthday party",
"date": "2024-02-14",
"location": "Alice's apartment",
"duration": 6
})
]
for participants, activity_info in activities:
social_network.add_e(participants, activity_info)
# Perform analysis first, then visualize
print("π Network Analysis Results:")
print("=" * 40)
# Analyze the network
print(f"π₯ Network size: {social_network.num_v} people, {social_network.num_e} activities")
# Find most social person
most_social = max(social_network.all_v, key=lambda v: social_network.degree_v(v))
print(f"π Most social person: {social_network.v(most_social)['name']} "
f"({social_network.degree_v(most_social)} activities)")
# Show activity count for all participants
print("\nπ₯ All participants' activity statistics:")
for person_id in social_network.all_v:
person_info = social_network.v(person_id)
activity_count = social_network.degree_v(person_id)
print(f" β’ {person_info['name']}: {activity_count} activities")
print("\n" + "=" * 40)
print("π¨ Starting visualization (press Ctrl+C to close)")
social_network.draw()
𧬠Example 2: Scientific Collaboration Network¶
Visualize research collaborations in computational biology:
from hyperdb import HypergraphDB
# Create research collaboration hypergraph
research_network = HypergraphDB()
# Add researchers as vertices
researchers = {
"dr_smith": {
"name": "Dr. Sarah Smith",
"field": "Bioinformatics",
"institution": "MIT",
"h_index": 45,
"experience": 15
},
"dr_jones": {
"name": "Dr. Michael Jones",
"field": "Machine Learning",
"institution": "Stanford",
"h_index": 38,
"experience": 12
},
"dr_garcia": {
"name": "Dr. Maria Garcia",
"field": "Genomics",
"institution": "Harvard",
"h_index": 52,
"experience": 18
},
"dr_chen": {
"name": "Dr. Wei Chen",
"field": "Systems Biology",
"institution": "UCSF",
"h_index": 41,
"experience": 14
},
"dr_taylor": {
"name": "Dr. James Taylor",
"field": "Computational Chemistry",
"institution": "Caltech",
"h_index": 36,
"experience": 10
}
}
for researcher_id, info in researchers.items():
research_network.add_v(researcher_id, info)
# Add research papers as hyperedges
publications = [
# Duo collaborations
(("dr_smith", "dr_jones"), {
"title": "Deep Learning for Protein Structure Prediction",
"journal": "Nature Biotechnology",
"year": 2023,
"citations": 127,
"impact_factor": 46.9
}),
# Trio collaborations
(("dr_garcia", "dr_chen", "dr_taylor"), {
"title": "Multi-omics Integration for Disease Prediction",
"journal": "Cell",
"year": 2023,
"citations": 98,
"impact_factor": 66.9
}),
# Large collaboration
(("dr_smith", "dr_jones", "dr_garcia", "dr_chen"), {
"title": "AI-Driven Drug Discovery Pipeline",
"journal": "Science",
"year": 2024,
"citations": 45,
"impact_factor": 56.9
}),
# Cross-institutional mega-collaboration
(("dr_smith", "dr_jones", "dr_garcia", "dr_chen", "dr_taylor"), {
"title": "The Future of Personalized Medicine",
"journal": "Nature Reviews Drug Discovery",
"year": 2024,
"citations": 23,
"impact_factor": 112.3
})
]
for authors, paper_info in publications:
research_network.add_e(authors, paper_info)
# Perform research impact analysis first
print("οΏ½ Research Collaboration Network Analysis:")
print("=" * 50)
# Basic network statistics
print(f"οΏ½ Network size: {research_network.num_v} researchers, {research_network.num_e} publications")
# Find most collaborative researcher
most_collaborative = max(research_network.all_v,
key=lambda v: research_network.degree_v(v))
researcher_info = research_network.v(most_collaborative)
print(f"π€ Most collaborative: {researcher_info['name']} "
f"({research_network.degree_v(most_collaborative)} papers)")
# Find highest-impact publication
highest_impact = max(research_network.all_e,
key=lambda e: research_network.e(e)['impact_factor'])
impact_factor = research_network.e(highest_impact)['impact_factor']
print(f"β Highest impact publication: {impact_factor} impact factor")
# Show detailed information for each researcher
print("\nπ¨βπ¬ Researcher collaboration statistics:")
for researcher_id in research_network.all_v:
info = research_network.v(researcher_id)
collab_count = research_network.degree_v(researcher_id)
print(f" β’ {info['name']} ({info['institution']})")
print(f" Field: {info['field']}, H-index: {info['h_index']}, Collaborations: {collab_count}")
print("\n" + "=" * 50)
print("π¬ Starting research network visualization (press Ctrl+C to close)")
research_network.draw()
π Example 3: E-commerce Purchase Patterns¶
Analyze customer purchase behaviors and product relationships:
from hyperdb import HypergraphDB
import random
# Create e-commerce hypergraph
ecommerce = HypergraphDB()
# Add products as vertices
products = {
"laptop": {"name": "Gaming Laptop", "category": "Electronics", "price": 1299.99, "rating": 4.5},
"mouse": {"name": "Wireless Mouse", "category": "Electronics", "price": 49.99, "rating": 4.3},
"keyboard": {"name": "Mechanical Keyboard", "category": "Electronics", "price": 129.99, "rating": 4.6},
"monitor": {"name": "4K Monitor", "category": "Electronics", "price": 399.99, "rating": 4.4},
"headset": {"name": "Gaming Headset", "category": "Electronics", "price": 89.99, "rating": 4.2},
"desk": {"name": "Standing Desk", "category": "Furniture", "price": 299.99, "rating": 4.1},
"chair": {"name": "Ergonomic Chair", "category": "Furniture", "price": 249.99, "rating": 4.7},
"lamp": {"name": "LED Desk Lamp", "category": "Furniture", "price": 79.99, "rating": 4.0}
}
for product_id, info in products.items():
ecommerce.add_v(product_id, info)
# Add purchase baskets as hyperedges
purchase_baskets = [
# Gaming setup purchase
(("laptop", "mouse", "keyboard", "headset"), {
"customer_id": "cust_001",
"purchase_date": "2024-01-15",
"total_amount": 1569.96,
"customer_type": "Gaming Enthusiast"
}),
# Office setup purchase
(("monitor", "desk", "chair", "lamp"), {
"customer_id": "cust_002",
"purchase_date": "2024-01-18",
"total_amount": 929.96,
"customer_type": "Remote Worker"
}),
# Minimal gaming setup
(("mouse", "keyboard", "headset"), {
"customer_id": "cust_003",
"purchase_date": "2024-01-20",
"total_amount": 269.97,
"customer_type": "Budget Gamer"
}),
# Luxury workspace
(("laptop", "monitor", "desk", "chair", "lamp"), {
"customer_id": "cust_004",
"purchase_date": "2024-01-25",
"total_amount": 2229.95,
"customer_type": "Professional"
}),
# Accessories only
(("mouse", "lamp"), {
"customer_id": "cust_005",
"purchase_date": "2024-01-28",
"total_amount": 129.98,
"customer_type": "Casual Buyer"
})
]
for products_in_basket, purchase_info in purchase_baskets:
ecommerce.add_e(products_in_basket, purchase_info)
# Perform market basket analysis first
print("οΏ½ E-commerce Purchase Pattern Analysis:")
print("=" * 45)
# Basic statistics
print(f"ποΈ Store overview: {ecommerce.num_v} products, {ecommerce.num_e} purchases")
# Find most popular product
most_popular = max(ecommerce.all_v, key=lambda v: ecommerce.degree_v(v))
product_info = ecommerce.v(most_popular)
print(f"π Most popular product: {product_info['name']} "
f"({ecommerce.degree_v(most_popular)} purchases)")
# Find largest purchase
largest_purchase = max(ecommerce.all_e, key=lambda e: len(ecommerce.e_v(e)))
num_items = len(ecommerce.e_v(largest_purchase))
purchase_info = ecommerce.e(largest_purchase)
print(f"π° Largest purchase: {num_items} items, ${purchase_info['total_amount']:.2f}")
# Show product popularity ranking
print(f"\nπ Product popularity ranking:")
products_by_popularity = sorted(ecommerce.all_v,
key=lambda v: ecommerce.degree_v(v),
reverse=True)
for i, product_id in enumerate(products_by_popularity, 1):
info = ecommerce.v(product_id)
purchases = ecommerce.degree_v(product_id)
print(f" {i}. {info['name']} - {purchases} purchases (${info['price']})")
# Analyze customer types
print(f"\nπ₯ Customer type analysis:")
customer_types = {}
for edge_id in ecommerce.all_e:
edge_data = ecommerce.e(edge_id)
customer_type = edge_data.get('customer_type', 'Unknown')
if customer_type not in customer_types:
customer_types[customer_type] = 0
customer_types[customer_type] += 1
for customer_type, count in customer_types.items():
print(f" β’ {customer_type}: {count} purchases")
print("\n" + "=" * 45)
print("π Starting purchase pattern visualization (press Ctrl+C to close)")
ecommerce.draw()
π¨ Visualization Customization Tips¶
1. Color Coding by Properties¶
The visualization automatically uses different colors for different types of vertices and hyperedges based on their properties.
2. Size Representation¶
- Vertex size: Often represents degree (number of connections)
- Hyperedge thickness: Represents the number of vertices it connects
3. Interactive Features¶
- Hover: View detailed information about vertices and hyperedges
- Click: Select elements to highlight related components
- Drag: Rearrange the layout for better viewing
- Zoom: Use mouse wheel to zoom in/out
4. Layout Algorithms¶
The visualization uses force-directed layout by default, which: - Groups related vertices together - Minimizes edge crossings - Creates aesthetically pleasing arrangements
5. Cross-Platform Compatibility π¶
Windows Users Note¶
On Windows systems, we've optimized Ctrl+C handling for the draw() function:
# Basic usage (blocking mode)
hg.draw() # Press Ctrl+C to stop server
# Non-blocking mode (recommended for scripts and automation)
viewer = hg.draw(blocking=False)
# Do other work...
viewer.stop_server() # Manually stop server
Platform Differences¶
| Operating System | Ctrl+C Behavior | Recommended Usage |
|---|---|---|
| Windows | β Works well after optimization | Both modes available |
| macOS/Linux | β Native support excellent | Default blocking mode |
Usage Recommendations¶
# 1. Interactive exploration (recommended blocking mode)
hg.draw(port=8080, blocking=True)
# 2. Script automation (recommended non-blocking mode)
viewer = hg.draw(port=8080, blocking=False)
# Perform other analysis...
time.sleep(30) # Give users time to view
viewer.stop_server()
# 3. Jupyter Notebook usage
viewer = hg.draw(blocking=False) # Don't block cell execution
π Analysis Through Visualization¶
Identifying Patterns¶
- Clusters: Groups of tightly connected vertices
- Hubs: Vertices with many connections (high degree)
- Bridges: Hyperedges that connect different clusters
- Outliers: Isolated or rarely connected vertices
Network Metrics Visualization¶
# Example: Analyze network centrality through visualization
def analyze_network_visually(hg):
print("π― Network Analysis:")
# Degree distribution
degrees = [hg.degree_v(v) for v in hg.all_v]
print(f"π Average degree: {sum(degrees)/len(degrees):.2f}")
# Hub identification
hubs = [v for v in hg.all_v if hg.degree_v(v) > sum(degrees)/len(degrees)]
print(f"π Network hubs: {len(hubs)} vertices")
# Hyperedge size distribution
edge_sizes = [hg.degree_e(e) for e in hg.all_e]
print(f"π Average hyperedge size: {sum(edge_sizes)/len(edge_sizes):.2f}")
# Visualize with analysis
hg.draw()
# Apply to any of the above examples
analyze_network_visually(social_network)
π Advanced Visualization Techniques¶
Dynamic Visualization¶
For time-series data, you can create multiple snapshots:
# Example: Evolving social network
def create_network_snapshots(base_network, time_periods):
snapshots = []
for period in time_periods:
# Create filtered network for each time period
period_network = HypergraphDB()
# Add vertices (people don't change)
for v in base_network.all_v:
period_network.add_v(v, base_network.v(v))
# Add only hyperedges from this time period
for e in base_network.all_e:
edge_data = base_network.e(e)
if edge_data.get('date', '') >= period['start'] and edge_data.get('date', '') <= period['end']:
period_network.add_e(base_network.e_v(e), edge_data)
snapshots.append((period['name'], period_network))
return snapshots
# Create quarterly snapshots
quarters = [
{"name": "Q1 2024", "start": "2024-01-01", "end": "2024-03-31"},
{"name": "Q2 2024", "start": "2024-04-01", "end": "2024-06-30"}
]
# Visualize evolution
for quarter_name, network in create_network_snapshots(social_network, quarters):
print(f"π
{quarter_name}:")
network.draw()
This visualization approach helps you understand the structure and evolution of complex relationships in your hypergraph data!