Scikit-network-13：Visualization
Visualization
Graph
使用SVG可视化图
from IPython.display import SVG
import numpy as np
from sknetwork.data import karate_club, painters, movie_actor, load_netset
from sknetwork.visualization import svg_graph, svg_digraph, svg_bigraph
图
graph = karate_club(metadata=True)
adjacency = graph.adjacency
position = graph.position
labels = graph.labels
# graph
image = svg_graph(adjacency, position, labels=labels)
SVG(image)
# export
image = svg_graph(adjacency, position, labels=labels, filename='karate_club')
# adding names
image = svg_graph(adjacency, position, names=np.arange(34), name_position='below', labels=labels)
SVG(image)
# node size
image = svg_graph(
adjacency, position, labels=labels, display_node_weight=True)
SVG(image)
# scores (here = degrees) 度数
degrees = adjacency.dot(np.ones(adjacency.shape[0]))
image = svg_graph(adjacency, position, scores=degrees)
SVG(image)
# seeds (here 2 nodes of highest degrees)
seeds = list(np.argsort(-degrees)[:2])
image = svg_graph(adjacency, position, labels=labels, seeds=seeds)
SVG(image)
weights = adjacency.dot(np.ones(adjacency.shape[0]))
image = svg_graph(adjacency, position, scores=np.log(weights), node_order=np.argsort(weights),
node_size_min=2, node_size_max=10, height=400, width=800,
display_node_weight=True, display_edges=False)
SVG(image)
有向图
graph = painters(metadata=True)
adjacency = graph.adjacency
names = graph.names
position = graph.position
image = svg_digraph(adjacency, position, names, name_position='above')
SVG(image)
二部图
graph = movie_actor(metadata=True)
biadjacency = graph.biadjacency
names_row = graph.names_row
names_col = graph.names_col
# default layout
image = svg_bigraph(biadjacency, names_row, names_col, color_row='blue', color_col='red')
SVG(image)
# keep original order of rows and columns
image = svg_bigraph(biadjacency, names_row, names_col=names_col, color_row='blue', color_col='red',
reorder=False)
SVG(image)
Path
路径的可视化
from IPython.display import SVG
from sknetwork.data import house, cyclic_digraph, star_wars
from sknetwork.visualization import svg_graph, svg_digraph, svg_bigraph
图
graph = house(metadata=True)
adjacency = graph.adjacency
position = graph.position
path = [(0, 1), (1, 2), (2, 3)]
edge_labels = [(*edge, 0) for edge in path]
edge_labels
# [(0, 1, 0), (1, 2, 0), (2, 3, 0)]
image = svg_graph(adjacency, position, edge_width=3, edge_labels=edge_labels)
SVG(image)
image = svg_graph(None, position, edge_width=3, edge_labels=edge_labels)
SVG(image)
有向图
graph = cyclic_digraph(10, metadata=True)
adjacency = graph.adjacency
position = graph.position
paths = [[(0, 1), (1, 2), (2, 3)], [(6, 7), (7, 8)]]
edge_labels = [(*edge, label) for label, path in enumerate(paths) for edge in path]
edge_labels
# [(0, 1, 0), (1, 2, 0), (2, 3, 0), (6, 7, 1), (7, 8, 1)]
image = svg_digraph(adjacency, position, width=200, height=None, edge_width=3, edge_labels=edge_labels)
SVG(image)
二部图
graph = star_wars(True)
biadjacency = graph.biadjacency
names_row = graph.names_row
names_col = graph.names_col3
path = [(0, 1), (2, 1)]
edge_labels = [(*edge, 0) for edge in path]
image = svg_bigraph(biadjacency, names_row=names_row, names_col=names_col, edge_width=3, edge_labels=edge_labels)
SVG(image)
Dendrograms
树状图可视化
from IPython.display import SVG
import numpy as np
from sknetwork.data import karate_club, painters, movie_actor
from sknetwork.hierarchy import Paris
from sknetwork.visualization import svg_graph, svg_digraph, svg_bigraph
from sknetwork.visualization import svg_dendrogram
图
graph = karate_club(metadata=True)
adjacency = graph.adjacency
position = graph.position
labels = graph.labels
# graph
image = svg_graph(adjacency, position, labels=labels)
SVG(image)
# hierarchical clustering， 层次聚类
paris = Paris()
dendrogram = paris.fit_transform(adjacency)
dendrogram
# visualization
image = svg_dendrogram(dendrogram)
SVG(image)
# add names, set colors
n = adjacency.shape[0]
image = svg_dendrogram(dendrogram, names=np.arange(n), n_clusters=5, color='gray')
SVG(image)
# export
svg_dendrogram(dendrogram, filename='dendrogram_karate_club') # save to local
SVG('./dendrogram_karate_club.svg')
有向图
graph = painters(metadata=True)
adjacency = graph.adjacency
names = graph.names
position = graph.position
# graph
image = svg_digraph(adjacency, position, names)
SVG(image)
# hierarchical clustering
paris = Paris()
dendrogram = paris.fit_transform(adjacency)
# visualization
image = svg_dendrogram(dendrogram, names, n_clusters=3, rotate=True)
SVG(image)
二部图
graph = movie_actor(metadata=True)
biadjacency = graph.biadjacency
names_row = graph.names_row
names_col = graph.names_col
# graph
image = svg_bigraph(biadjacency, names_row, names_col)
SVG(image)
# hierarchical clustering
paris = Paris()
paris.fit(biadjacency)
dendrogram_row = paris.dendrogram_row_
dendrogram_col = paris.dendrogram_col_
dendrogram_full = paris.dendrogram_full_
# visualization
image = svg_dendrogram(dendrogram_row, names_row, n_clusters=3, rotate=True)
SVG(image)
image = svg_dendrogram(dendrogram_col, names_col, n_clusters=3, rotate=True)
SVG(image)
Pie-chart nodes
饼图可视化
from IPython.display import SVG
from scipy import sparse
from sknetwork.data import bow_tie, karate_club, painters
from sknetwork.visualization import svg_graph, svg_digraph
from sknetwork.clustering import Louvain
图
graph = bow_tie(metadata=True)
adjacency = graph.adjacency
position = graph.position
# probabilities
probs = [.5, 0, 0, 1, 1]
membership = sparse.csr_matrix([[p, 1-p] for p in probs])
image = svg_graph(adjacency, position, membership=membership, node_size=10)
SVG(image)
graph = karate_club(True)
adjacency = graph.adjacency
position = graph.position
# soft clustering
louvain = Louvain()
louvain.fit(adjacency)
membership = louvain.membership_
image = svg_graph(adjacency, position, membership=membership)
SVG(image)
有向图
graph = painters(True)
adjacency = graph.adjacency
names = graph.names
positions = graph.position
# soft clustering
louvain = Louvain()
louvain.fit(adjacency)
membership = louvain.membership_
image = svg_digraph(adjacency, names=names, position=positions, membership=membership, node_size=10)
SVG(image)
