SNAP¶

This Jupyter notebook provides an example of using the Python packages gravis and SNAP (=Stanford Network Analysis Platform). The .ipynb file can be found here.

References¶

SNAP website
- Snap.py
  - Quick Introduction
  - Documentation
    - Tutorial
    - Reference manual
- Datasets
- Tutorials (Teaching)

Installation¶

With pip: pip install snap-stanford
With conda: Not available except for MacOS

Import¶

[1]:

import os

import snap

import gravis as gv

Graph construction¶

1) Manual graph construction¶

Tutorial: Graph creation
API reference: Graph and network classes

2) Algorithmic graph construction¶

[13]:

num_nodes = 20
num_edges = 10

[14]:

g = snap.GenForestFire(num_nodes, 0.35, 0.35)

# A random Erdos-Renyi directed graph
g = snap.GenRndGnm(snap.PNGraph, num_nodes, num_edges)

# A Preferential Attachment graph with out-degree 3
g = snap.GenPrefAttach(num_nodes, 3)

3) Graph loading from an internal collection¶

[15]:

# TODO

4) Graph import and export¶

Import¶

Tutorial: Saving and Loading Graphs

[16]:

filepath = os.path.join('data', 'snap_graph.net')
g = snap.LoadPajek(snap.PNGraph, filepath)

# TODO

Export¶

Tutorial: Saving and Loading Graphs

[17]:

filepath = os.path.join('data', 'snap_graph.net')
snap.SavePajek(g, filepath)

# Other methods
# snap.SaveEdgeList(g, 'snap_graph_edge_list.txt')
# TODO

Basic graph inspection¶

1) Graph and its properties¶

[18]:

print('Type:', type(g))
print('Directed:', )
print('Number of nodes:', g.GetNodes())
print('Number of edges:', g.GetEdges())
print('Well-formed:', g.IsOk())
print('Approximate graph diameter:', snap.GetBfsFullDiam(g, 10))
print('Number of triads:', snap.GetTriads(g))
print('Clustering coefficient', snap.GetClustCf(g))

Type: <class 'snap.PNGraph'>
Directed:
Number of nodes: 50
Number of edges: 100
Well-formed: True
Approximate graph diameter: 5
Number of triads: 9
Clustering coefficient 0.08790476190476189

2) Nodes and their properties¶

[19]:

#g.AddIntAttrN("NValInt", 0)
for node in g.Nodes():
    print('Class:', type(node))
    print("Id:", node.GetId())
    print("In degree:", node.GetInDeg())
    print("Out degree:", node.GetOutDeg())
    break
    #g.AddIntAttrDatN(node_id, 42, "NValInt")

Class: <class 'snap.TNGraphNodeI'>
Id: 1
In degree: 0
Out degree: 2

3) Edges and their properties¶

[20]:

for edge in g.Edges():
    print('Class:', type(edge))
    print("Id:", edge.GetId())
    print("Source node id:", edge.GetSrcNId())
    print("Target node id:", edge.GetDstNId())
    break

Class: <class 'snap.TNGraphEdgeI'>
Id: (1, 4)
Source node id: 1
Target node id: 4

Calculating graph measures and metrics¶

1) Quantitative measures¶

[21]:

snap.GetClustCf(g)

[21]:

0.08790476190476189

2) Structure inference¶

Community detection¶

Docs: Community Detection

[22]:

g = snap.GenRndGnm(snap.PUNGraph, 50, 100)

CmtyV = snap.TCnComV()
modularity = snap.CommunityGirvanNewman(g, CmtyV)

CmtyV = snap.TCnComV()
modularity = snap.CommunityCNM(g, CmtyV)

Other kinds of groups¶

[23]:

g = snap.GenRndGnm(snap.PUNGraph, 50, 100)

snap.GetTriads(g)

[23]:

Graph visualization¶

[24]:

# TODO

SNAP¶

References¶

Installation¶

Import¶

Quick start¶

Example 1¶

Graph construction¶

1) Manual graph construction¶

1.a) TUNGraph¶

1.b) TNGraph¶

1.c) TUndirNet (similar to TUNGraph)¶

1.d) TDirNet (similar to TNGraph)¶

1.e) TNEANet¶

Assign attributes to a created graph¶

1) Graph attributes¶

2) Node attributes¶

3) Edge attributes¶

2) Algorithmic graph construction¶

3) Graph loading from an internal collection¶

4) Graph import and export¶

Import¶

Export¶

Basic graph inspection¶

1) Graph and its properties¶

2) Nodes and their properties¶

3) Edges and their properties¶

Calculating graph measures and metrics¶

1) Quantitative measures¶

2) Structure inference¶

Community detection¶

Other kinds of groups¶

Graph visualization¶