Plot a graph with Matplotlib

plot_graph.py

import matplotlib as mpl
import matplotlib.pyplot as plt
%matplotlib inline
import networkx as nx


def plot_graph(df, directed=False):
    graph_engine = nx.DiGraph() if directed else None
    G = nx.from_pandas_edgelist(df, source='source', target='target', edge_attr=True, create_using=graph_engine)
    costs = G.edges.data('cost')

    pos = nx.layout.spring_layout(G)
    pos_labels = nx.layout.spring_layout(G, center=(0.05,0.05))

    node_sizes = [10 for i in range(len(G))]
    M = G.number_of_edges()
    edge_colors = [x[2] for x in G.edges.data('cost')]


    nodes = nx.draw_networkx_nodes(G, pos, node_size=node_sizes, node_color='blue')
    labels = nx.draw_networkx_labels(G, pos=pos_labels)
    edges = nx.draw_networkx_edges(G, pos, node_size=node_sizes, arrowstyle='->',
                                   arrowsize=10, edge_color=edge_colors,
                                   edge_vmin=0., edge_cmap=plt.cm.Blues, width=2)

    # set alpha value for each edge
    if directed:
        edge_alphas = [0.5 for i in range(M)]
        for i in range(M):
            edges[i].set_alpha(edge_alphas[i])

        pc = mpl.collections.PatchCollection(edges, cmap=plt.cm.Blues)
        pc.set_array(edge_colors)
        plt.colorbar(pc)

    ax = plt.gca()
    ax.set_axis_off()
    plt.show()
    
df = pd.DataFrame({"source": [1, 2, 3], "target": [2, 3, 1], "cost": [4, 5, 6]})
plot_graph(df)