Plot degree distribution in log-log scale

Question

I have a graph in networkx and im trying to plot the degree distribution. So i wrote the following code:

g = nx.read_edgelist('graph.txt', create_using= nx.Graph(), nodetype=int)

print(nx.info(g))
numOfNodes = []
for i in range(0, g.number_of_nodes()):
    numOfNodes.append(i)

s = sorted(g.degree, key=lambda x: x[1], reverse=True)
degrees = sorted([x[0] for x in s])
frequency = sorted([x[1] for x in s])
plt.loglog(numOfNodes, degrees)
plt.xlabel("Degree")
plt.ylabel("No. of nodes (Frequency)")
plt.title("Degree distribution")
plt.show()

I expected to get a diagram like this:

but im getting something much different:

What is wrong with my plot of degree distribution? I can't find what im doing wrong.

Answer 1

These are the two ways I know to obtain the degree distribution:

1. Using the networkX degree_histogram function( docs ):

Here's an example on how to use:

def plot_degree_histogram(g, normalized=True):
    print("Creating histogram...")
    aux_y = nx.degree_histogram(g)
    
    aux_x = np.arange(0,len(aux_y)).tolist()
    
    n_nodes = g.number_of_nodes()
    
    if normalized:
        for i in range(len(aux_y)):
            aux_y[i] = aux_y[i]/n_nodes
    
    return aux_x, aux_y

2. In case you are using weights, follow this example from networkX documentation:

def plot_degree_histogram(g, normalized=True, weight=None):
    
    degree_sequence = sorted([d for n, d in g.degree(weight=weight)], reverse=True)  # degree sequence
    degreeCount = collections.Counter(degree_sequence)
    aux_x, aux_y = zip(*degreeCount.items())

    n_nodes = g.number_of_nodes()
    aux_y = list(aux_y)
    if normalized:
        for i in range(len(aux_y)):
            aux_y[i] = aux_y[i]/n_nodes
    
    return aux_x, aux_y

Plotting

Simple as:

plt.title('\nDistribution Of Node Linkages (log-log scale)')
plt.xlabel('Degree\n(log scale)')
plt.ylabel('Number of Nodes\n(log scale)')
plt.xscale("log")
plt.yscale("log")
plt.plot(aux_x, aux_y, 'o')

About you implementation:

I believe that your degrees list is valid. However numOfNodes is not what you are looking for, ie not the counting of nodes of each degree.
Hope the above solutions help!