mpld3.show() returns Object of type int is not JSON serializable

I want to create a network where you can hover over each label to read it interactively.

I am using jupyter lab, specs are: Selected Jupyter core packages...

IPython          : 7.6.1
ipykernel        : 5.1.1
ipywidgets       : 7.6.5
jupyter_client   : 7.0.6
jupyter_core     : 4.8.1
jupyter_server   : not installed
jupyterlab       : 1.0.2
nbclient         : not installed
nbconvert        : 5.5.0
nbformat         : 4.4.0
notebook         : 6.0.0
qtconsole        : 4.5.1
traitlets        : 4.3.2

When I run this code in a jupyter notebook:

import matplotlib.pyplot as plt
import numpy as np
import mpld3

fig, ax = plt.subplots(subplot_kw=dict(axisbg='#EEEEEE'))
N = 100

scatter = ax.scatter(np.random.normal(size=N),
                     np.random.normal(size=N),
                     c=np.random.random(size=N),
                     s=1000 * np.random.random(size=N),
                     alpha=0.3,
                     cmap=plt.cm.jet)
ax.grid(color='white', linestyle='solid')

ax.set_title("Scatter Plot (with tooltips!)", size=20)

labels = ['point {0}'.format(i + 1) for i in range(N)]
tooltip = mpld3.plugins.PointLabelTooltip(scatter, labels=labels)
mpld3.plugins.connect(fig, tooltip)

mpld3.show()

That I obtained from here, a new window opens, with interactive labels, as expected and identical to the example in the hyperlink.

My own data is:

index col_A
0     6840
1     6640
2      823
3    57019

index col_B
0     7431
1     5217
2     7431
3    57019

For a network, these are pairs of node labels like this:

col_A  col_B
6840   7431
6640   5217
823    7431
57019  57019

So the output network should have three clusters:

6840-7431-823
6640-5217
57019-57019

When I run this code, which is almost identical to the example code above:

import matplotlib.pyplot as plt
import numpy as np
import mpld3
import mplcursors



import networkx as nx
#G = nx.path_graph(4)
#pos = nx.spring_layout(G)

G = nx.from_pandas_edgelist(final_net,'col_A','col_B',['col_A', 'col_B'])
print(final_net['col_A'][0:10])
print(final_net['col_B'][0:10])

edge_labels = nx.get_edge_attributes(G, "Edge_label")
pos = nx.spring_layout(G)


fig, ax = plt.subplots(subplot_kw=dict(facecolor='#EEEEEE'))
scatter = nx.draw_networkx_nodes(G, pos, ax=ax)
nx.draw_networkx_edges(G, pos, ax=ax)

labels = G.nodes()
tooltip = mpld3.plugins.PointLabelTooltip(scatter, labels=labels)
mpld3.plugins.connect(fig, tooltip)
mplcursors.cursor(hover=True)

mpld3.show()

I do get the correct static image:

But I get an error:

TypeError: Object of type int is not JSON serializable

And the network doesn't open in a new window that I can interact with (ideally the interactive network would remain in jupyter anyway).

I changed the object types to string to see what happened with:

final_net['col_A'] = pd.to_numeric(final_net['col_A'])
final_net['col_B'] = pd.to_numeric(final_net['col_B'])

With the output:

col_A    int64
col_B    int64

But the error remains the same. When I remove the last line, mpld3.show() , the error disappears, so I just get a static image as an output, with no error, but no interactivity either.

I uninstalled and re-installed with conda as per here (which keeps the same error) and then I dumped to JSON as per here

by doing:

import json
import numpy as np

data = [[6840, 7431], [6640, 5217], [823, 7431],[57019,57019]]
final_net = pd.DataFrame(data, columns = ['col_A', 'col_B'])

class NumpyEncoder(json.JSONEncoder):
    """ Special json encoder for numpy types """
    def default(self, obj):
        if isinstance(obj, np.integer):
            return int(obj)
        elif isinstance(obj, np.floating):
            return float(obj)
        elif isinstance(obj, np.ndarray):
            return obj.tolist()
        return json.JSONEncoder.default(self, obj)

#dumped = json.dumps(final_net, cls=NumpyEncoder)

#with open(path, 'w') as f:
#    json.dump(dumped, f)
    
final_net['col_A'] = json.dumps(final_net['col_A'],cls=NumpyEncoder)
final_net['col_B'] = json.dumps(final_net['col_B'],cls=NumpyEncoder)

When I dump to json and then rerun my network code again, it outputs:

0    "{\"0\":6840,\"1\":6640,\"2\":823,\"3\":57019}"
1    "{\"0\":6840,\"1\":6640,\"2\":823,\"3\":57019}"
2    "{\"0\":6840,\"1\":6640,\"2\":823,\"3\":57019}"
3    "{\"0\":6840,\"1\":6640,\"2\":823,\"3\":57019}"
Name: Entrez Gene Interactor A, dtype: object
0    "{\"0\":7431,\"1\":5217,\"2\":7431,\"3\":57019}"
1    "{\"0\":7431,\"1\":5217,\"2\":7431,\"3\":57019}"
2    "{\"0\":7431,\"1\":5217,\"2\":7431,\"3\":57019}"
3    "{\"0\":7431,\"1\":5217,\"2\":7431,\"3\":57019}"

And this image (which is wrong), and no interactivity.

I'm wonder if someone could show me how to edit my code to make the interactive feature appear (ideally in a jupyter notebook, if not it's ok if it opens in a new window).

The problem seems to be that G.nodes() isn't a list of labels. You can get the node numbers or labels via converting it to a list (list(G.nodes())).

An updated version could look like:

import matplotlib.pyplot as plt
import networkx as nx
import pandas as pd
import numpy as np
import mpld3

final_net = pd.DataFrame({'col_A': [6840, 6640, 823, 57019],
                          'col_B': [7431, 5217, 7431, 57019]})
G = nx.from_pandas_edgelist(final_net, 'col_A', 'col_B', ['col_A', 'col_B'])
print(final_net['col_A'][0:10])
print(final_net['col_B'][0:10])

edge_labels = nx.get_edge_attributes(G, "Edge_label")
pos = nx.spring_layout(G)

fig, ax = plt.subplots(subplot_kw=dict(facecolor='#EEEEEE'))
scatter = nx.draw_networkx_nodes(G, pos, ax=ax)
nx.draw_networkx_edges(G, pos, ax=ax)

labels = list(G.nodes())
tooltip = mpld3.plugins.PointLabelTooltip(scatter, labels=labels)
mpld3.plugins.connect(fig, tooltip)

mpld3.show()