在 Python 中生成具有平行标记 Edges/Vertices 的有向图

Question

在 Python 中生成具有平行标记 Edges/Vertices 的有向图的最佳方法是什么，如下图所示？

我已经尝试过 networkx，但它不适用于平行边。

这是我用来生成图表数据的代码。

from forex_python.converter import CurrencyRates
import pandas as pd

chosen_currencies = ['GBP', 'EUR', 'USD']

c = CurrencyRates()

rates = []


for currency_index in range(len(chosen_currencies)):
    temp_list = []
    for second_index in range(len(chosen_currencies)):
        temp_list.append(c.get_rate(chosen_currencies[currency_index], chosen_currencies[second_index]))
    rates.append(temp_list)

df = (pd.DataFrame.from_records(rates)).transpose()
df.columns = chosen_currencies

Answer 1

您可以使用数据框通过 from_pandas_adjacency 将边直接读入 NetworkX 图中。为此，让我们将 dataframe 的索引设置为 chosen_currencies，以确保正确映射边缘。

from forex_python.converter import CurrencyRates
import pandas as pd

chosen_currencies = ['GBP', 'EUR', 'USD']

c = CurrencyRates()

rates = []


for currency_index in range(len(chosen_currencies)):
    temp_list = []
    for second_index in range(len(chosen_currencies)):
        temp_list.append(c.get_rate(chosen_currencies[currency_index], chosen_currencies[second_index]))
    rates.append(temp_list)

df = (pd.DataFrame.from_records(rates)).transpose()
df.columns = chosen_currencies
#   GBP         EUR     USD
#0  1.000000    0.83238 0.768233
#1  1.201374    1.00000 0.922935
#2  1.301689    1.08350 1.000000

现在设置索引

df.set_index([pd.Index(chosen_currencies)], inplace=True)
#       GBP         EUR     USD
#GBP    1.000000    0.83238 0.768233
#EUR    1.201374    1.00000 0.922935
#USD    1.301689    1.08350 1.000000

现在让我们创建图表

import networkx as nx
import matplotlib.pyplot as plt

G = nx.from_pandas_adjacency(df, create_using=nx.DiGraph)

# Set the figure size
plt.figure(figsize=(8,8))

# Get the edge labels and round them to 3 decimal places
# for more clarity while drawing
edge_labels = dict([((u,v), round(d['weight'], 3))
             for u,v,d in G.edges(data=True)])

# Get the layout
pos = nx.spring_layout(G)

# Draw edge labels and graph
nx.draw_networkx_edge_labels(G,pos,edge_labels=edge_labels,
                             label_pos=0.15, font_size=10)
nx.draw(G, pos, with_labels=True,
        connectionstyle='arc3, rad = 0.15',
        node_size=800)

plt.show()

注：箭头附近的数字为边权重

您还可以查看this Google Colab Notebook上面的工作示例。

您可以根据需要调整字体大小、标签位置等。

参考文献：

Connection Style
Draw Edge Labels
Drawing multiple edge labels

在 Python 中生成具有平行标记 Edges/Vertices 的有向图

Generating Directed Graph With Parallel Labelled Edges/Vertices in Python

python

currency

graph

directed-graph