从时间序列到复杂网络：可见图算法

# coding: utf-8
import networkx as nx
import matplotlib.pyplot as plt
from itertools import combinations

def visibility_graph(series):
visibility_graph_edges=[]
# convert list of magnitudes into list of tuples that hold the index
tseries = []
n = 1
for magnitude in series:
tseries.append( (n, magnitude ) )
n += 1

for a,b in combinations(tseries, 2):
# two points, maybe connect
(ta, ya) = a  #实现（1,2)（1,3）(1,4)(1,5)--(2,3)(2,4)(2,5)--(3,4)(3,5)--(4,5)任意两个边相互比较
(tb, yb) = b
connect = True
medium=tseries[ta:tb-1]#此处需要多留意，ta是1到k，而tseris是从0下标开始  所以此处不能不是[ta+1:tb]

for tc,yc in medium:
#print yc,(yb + (ya - yb) * ( float(tb - tc) / (tb - ta) ))#一定要float(tb-tc)/(tb-ta)因为计算机里1/2为0,1.0/2才为0.5
if yc > yb + (ya - yb) * ( float(tb - tc) / (tb - ta) ):
connect =False
if connect:
visibility_graph_edges.append((ta,tb))
return visibility_graph_edges

def DrawVisibilityGraph(graphEdges):

G=nx.Graph()  #G=nx.Graph()无向图    G=nx.DiGraph()有向图
#add the edge to graph G
for i in graphEdges:
G.add_edge(i[0],i[1])

pos=nx.spring_layout(G)
nx.draw_networkx_nodes(G,pos,node_size=700)
nx.draw_networkx_edges(G,pos,width=5)
nx.draw_networkx_labels(G,pos,font_size=20,font_family='sans-serif')
#plt.savefig("test.png")   #save the picture
plt.show()

if __name__=="__main__":
#time series
series = [0.87,0.49,0.36,0.83,0.87]
#get the graph edges
graphEdges=visibility_graph(series)
#draw the visibility graph
DrawVisibilityGraph(graphEdges)