Coverage for /pythoncovmergedfiles/medio/medio/usr/local/lib/python3.9/dist-packages/networkx/generators/time_series.py: 36%

1""" 

2Time Series Graphs 

3""" 

4import itertools 

5 

6import networkx as nx 

7 

8__all__ = ["visibility_graph"] 

9 

10 

11@nx._dispatch(graphs=None) 

12def visibility_graph(series): 

13 """ 

14 Return a Visibility Graph of an input Time Series. 

15 

16 A visibility graph converts a time series into a graph. The constructed graph 

17 uses integer nodes to indicate which event in the series the node represents. 

18 Edges are formed as follows: consider a bar plot of the series and view that 

19 as a side view of a landscape with a node at the top of each bar. An edge 

20 means that the nodes can be connected by a straight "line-of-sight" without 

21 being obscured by any bars between the nodes. 

22 

23 The resulting graph inherits several properties of the series in its structure. 

24 Thereby, periodic series convert into regular graphs, random series convert 

25 into random graphs, and fractal series convert into scale-free networks [1]_. 

26 

27 Parameters 

28 ---------- 

29 series : Sequence[Number] 

30 A Time Series sequence (iterable and sliceable) of numeric values 

31 representing times. 

32 

33 Returns 

34 ------- 

35 NetworkX Graph 

36 The Visibility Graph of the input series 

37 

38 Examples 

39 -------- 

40 >>> series_list = [range(10), [2, 1, 3, 2, 1, 3, 2, 1, 3, 2, 1, 3]] 

41 >>> for s in series_list: 

42 ... g = nx.visibility_graph(s) 

43 ... print(g) 

44 Graph with 10 nodes and 9 edges 

45 Graph with 12 nodes and 18 edges 

46 

47 References 

48 ---------- 

49 .. [1] Lacasa, Lucas, Bartolo Luque, Fernando Ballesteros, Jordi Luque, and Juan Carlos Nuno. 

50 "From time series to complex networks: The visibility graph." Proceedings of the 

51 National Academy of Sciences 105, no. 13 (2008): 4972-4975. 

52 https://www.pnas.org/doi/10.1073/pnas.0709247105 

53 """ 

54 

55 # Sequential values are always connected 

56 G = nx.path_graph(len(series)) 

57 nx.set_node_attributes(G, dict(enumerate(series)), "value") 

58 

59 # Check all combinations of nodes n series 

60 for (n1, t1), (n2, t2) in itertools.combinations(enumerate(series), 2): 

61 # check if any value between obstructs line of sight 

62 slope = (t2 - t1) / (n2 - n1) 

63 offset = t2 - slope * n2 

64 

65 obstructed = any( 

66 t >= slope * n + offset 

67 for n, t in enumerate(series[n1 + 1 : n2], start=n1 + 1) 

68 ) 

69 

70 if not obstructed: 

71 G.add_edge(n1, n2) 

72 

73 return G