属性错误：模块“networkx”没有属性“from_scipy_sparse_matrix”

Question

AttributeError                            Traceback (most recent call last)
    <ipython-input-85-6e1936149908> in <cell line: 2>()
          1 fig, ax = plt.subplots(figsize=(8, 8))
    ----> 2 ax = draw_graph(A, ax=ax, size_factor=1)
    
    <ipython-input-83-384ee4aa675e> in draw_graph(A, m, ax, spring_layout, size_factor)
         26     assert m**2 == A.shape[0] == A.shape[1]
         27     # Create the nx.Graph object
    ---> 28     G = nx.from_scipy_sparse_matrix(A)
         29     print('Number of nodes: %d; Number of edges: %d' % \
         30           (G.number_of_nodes(), G.number_of_edges()))
    
    AttributeError: module 'networkx' has no attribute 'from_scipy_sparse_matrix'

以下错误发生：

    fig, ax = plt.subplots(figsize=(8, 8))
    ax = draw_graph(A, ax=ax, size_factor=1)

定义图表的代码：

   def grid_graph(m, k=8, corners=False):
    '''
    To create adjacency matrix as per Defferrard et al. 2016
    '''
    z = graph.grid(m)
    dist, idx = graph.distance_sklearn_metrics(z, k=k, metric='euclidean')
    A = graph.adjacency(dist, idx)

    # Connections are only vertical or horizontal on the grid.
    # Corner vertices are connected to 2 neightbors only.
    if corners:
        import scipy.sparse
        A = A.toarray()
        A[A < A.max()/1.5] = 0
        A = scipy.sparse.csr_matrix(A)
        print('{} edges'.format(A.nnz))

    print("{} > {} edges".format(A.nnz//2, k*m**2//2))
    return A


    def draw_graph(A, m=28, ax=None, spring_layout=False, size_factor=10):
        '''Draw the graph given adjacency matrix(A),
        optionally with spring_layout.
        '''
        assert m**2 == A.shape[0] == A.shape[1]
        # Create the nx.Graph object
        G = nx.from_scipy_sparse_matrix(A)
        print('Number of nodes: %d; Number of edges: %d' % \
              (G.number_of_nodes(), G.number_of_edges()))
        grid_coords = graph.grid(m)
    
        if spring_layout:
            # remove nodes without edges
            nodes_without_edges = [n for n, k in  G.degree() if k == 0]
            G.remove_nodes_from(nodes_without_edges)
            print('After removing nodes without edges:')
            print('Number of nodes: %d; Number of edges: %d' % \
                  (G.number_of_nodes(), G.number_of_edges()))
        
        z = graph.grid(m)
        
        # initial positions
        pos = {n: z[n] for n in G.nodes()} 
        
        if spring_layout:
            pos = nx.spring_layout(G, 
                                   pos=pos,
                                   iterations=200)
        
        ax = nx.draw(G, pos,
                     node_size=[G.degree(n) * size_factor for n in G.nodes()],
                     ax=ax
                    )
        return ax

Answer 1

NetworkX 2.6 版本中引入了 from_scipy_sparse_matrix 函数。如果您使用的是旧版本的 NetworkX，您将无法使用此功能。要解决此问题，您可以将 NetworkX 版本升级到 2.6

pip uninstall networkx -y
pip install networkx==2.6.0