Plot network from large pandas dataframe using networkx

Question

I am writing on jupyter a program for the statistical validation of a network, the final product is a large pandas dataframe 5053x5053:

import pandas as pd
network = pd.DataFrame (data = app, index = products, columns = products)

app is a binary matrix where if app[i,j] = 1 the product i is linked to the product j. I would like to plot the network, and I just learned that it is possible using networkx (and sometimes other tools like cytoscape). Since the amount of data is large I have no clue on how to procede. Which kind of representation is the best and how can I obtain a readable plot? I have tried to write down some basic code, but results are quite disappointing:

import networkx as nx
import matplotlib.pyplot as plt

G = nx.Graph()
G = nx.from_pandas_edgelist(network)
nx.draw_random(G)

Furthermore I have a vector of 212 green products serial numbers (indexes and columns of the dataframe) that if possible I would like to draw of a different color on the same plot.

Edit: I used the code and it works better than my try, but it is still not a readble graph.

G = nx.from_numpy_matrix(gg)
G = nx.relabel_nodes(graph, dict(enumerate(greenxgreen.columns)))
nx.draw(G)

CypherX · Accepted Answer · 2020-07-16 18:07:44Z

1

Solution

I have used the dataframe (df) from the Dummy Data section below. This would give you a basic network-diagram. I would encourage you to dig in further in the documentation (see References section).

The nx.draw_random() creates a random arrangement. You have a lot of nodes. And so it will create a clutter. You might want to select a subset of the dataframe that has a certain number of connections at least and plot them instead to reduce the clutter.

# G = graph
G = nx.from_numpy_matrix(df.values)
G = nx.relabel_nodes(G, dict(enumerate(df.columns)))
# nx.draw_spectral(G)
# nx.draw_random(G)
# nx.draw_circular(G)
nx.draw(G)

To draw the whole network, including the labels, edges, etc. use nx.draw_networkx().

nx.draw_networkx(G)

Dummy Data

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

import networkx as nx

%matplotlib inline

## To randomly generate array: a
#  Uncomment the following three lines
# seed = 0
# np.random.seed(seed=seed)
# a = (np.random.rand(25).reshape(5,5) >= 0.5).astype(int)

## To use a fixed representation of array: a
a = np.array([
    [1, 1, 1, 1, 0],
    [1, 0, 1, 1, 0],
    [1, 1, 1, 1, 0],
    [0, 0, 1, 1, 1],
    [1, 1, 0, 1, 0]
    ])

nodes = list('ABCDE')
df = pd.DataFrame(data=a, index=nodes, columns=nodes)
print(df)

References

edited Jul 16, 2020 at 18:07

answered Jul 15, 2020 at 0:56

CypherX

7,4034 gold badges29 silver badges39 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

CypherX Over a year ago

@Rodolfo-Moreschi Is this what you were looking for?

Rodolfo Over a year ago

Thanks, yes pretty much it is, I'll try right away, the only problem is that having a lot of nodes I don't know how to make a readable graph. I'll try and post the code and image.

Rodolfo Over a year ago

I've just added a sample of the graphs, is it a subnetwork smaller than the network itself (that is 5053x5053)

Rodolfo Over a year ago

Working on the parameters of the plot (nodes size and color, and edge color and width), but it took me one day to learn how to do it

Collectives™ on Stack Overflow

Plot network from large pandas dataframe using networkx

1 Answer 1

Solution

Dummy Data

References

4 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Solution

Dummy Data

References

4 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related