1

I am trying to draw a bipartite graph for my data set, which is like below:

source  target  weight
reduce  energy  25
reduce  consumption 25
energy  pennsylvania    4
energy  natural 4
consumption balancing   4

the code That I am trying to plot the graph is as below:

C_2021 = nx.Graph()
C_2021.add_nodes_from(df_final_2014['source'], bipartite=0)
C_2021.add_nodes_from(df_final_2014['target'], bipartite=1)
edges = df_final_2014[['source', 'target','weight']].apply(tuple, axis=1)
C_2021.add_weighted_edges_from(edges)

But when I check with the below code whether it is bipartite or not, I get the "False" feedback.

nx.is_bipartite(C_2021)

Could you please advise what the issue is?

The previous issue is resolved, but when I want to plot the bipartite graph with the below steps, I do not get a proper result. If someone could help me, I will be appreciated it:

top_nodes_2021 = set(n for n,d in C_2021.nodes(data=True) if d['bipartite']==0)
top_nodes_2021

the output of the above is:

{'reduce'}

bottom_nodes_2021 = set(C_2021) - top_nodes_2021
bottom_nodes_2021

the output of the above is:

{'balancing', 'consumption', 'energy', 'natural', 'pennsylvania '}

then plot it by:

pos = nx.bipartite_layout(C_2021,top_nodes_2021)
plt.figure(figsize=[8,6])
# Pass that layout to nx.draw
nx.draw(C_2021,pos,node_color='#A0CBE2',edge_color='black',width=0.2,
     edge_cmap=plt.cm.Blues,with_labels=True)

and the result is:

enter image description here

code_lover
  • 25
  • 5

1 Answers1

1

It works for me using your code. nx.is_bipartite(C_2021) returns true. Check the example below:

import sys
if sys.version_info[0] < 3: 
    from StringIO import StringIO
else:
    from io import StringIO

import pandas as pd

data = StringIO('''source;target;weight
reduce;energy;25
reduce;consumption;25
energy;pennsylvania ;4
energy;natural;4
consumption;balancing;4
''')

df_final_2014 = pd.read_csv(data, sep=";")

C_2021 = nx.Graph()
C_2021.add_nodes_from(df_final_2014['source'], bipartite=0)
C_2021.add_nodes_from(df_final_2014['target'], bipartite=1)
edges = df_final_2014[['source', 'target','weight']].apply(tuple, axis=1)
C_2021.add_weighted_edges_from(edges)

nx.is_bipartite(C_2021)

Finally to draw them get the bipartite sets. The data you passed during the creation is false (i.g. bipartite=0 and bipartite=1).

Use the following commands:

from networkx.algorithms import bipartite
top_nodes_2021, bottom_nodes_2021 = bipartite.sets(C_2021)
pos = nx.bipartite_layout(C_2021, top_nodes_2021)

plt.figure(figsize=[8,6])
# Pass that layout to nx.draw
nx.draw(C_2021,pos,node_color='#A0CBE2',edge_color='black',width=0.2,
     edge_cmap=plt.cm.Blues,with_labels=True)

With the following result: enter image description here

AveragePythonEnjoyer
  • 1,115
  • 3
  • 14
  • Thank you, it is resolved. but after that, I want to draw the top and bottom nodes from the data to plot the bipartite graph, but for the top nodes, I always get one node which is: {'reduce'} is this code correct? top_nodes_2021 = {n for n, d in C_2021.nodes(data=True) if d['bipartite']==0} top_nodes_2021 – code_lover Jul 20 '22 at 14:24
  • You have got an edge from 'reduce' to 'energy' - so the selection on `d['bipartite']==0` will only return reduce. – AveragePythonEnjoyer Jul 20 '22 at 15:07
  • Thank you. could you please advise how I can clean my data set and draw only a bipartite graph? I have deleted the edge from 'reduce' to 'energy'; the output now is {'energy', 'reduce'}. I think I have to do something with my dataset to draw the bipartite graph. – code_lover Jul 20 '22 at 15:20
  • I do not understand what your goal is in the end. Could you explain it a bit better? If you want to draw the network use: `nx.draw(C_2021)`. – AveragePythonEnjoyer Jul 20 '22 at 16:21
  • the issue has been updated. I really appreciated your support. – code_lover Jul 21 '22 at 04:43
  • Thanks a lot for your help. when I apply this approach to the complete dataset, in return for nx.is_bipartite(C_2021), I got false. Is it confirmed that the graph from the full dataset is not bipartite at all? – code_lover Jul 21 '22 at 13:05
  • Yes, if nx.is_bipartite() returns false then the graph is not bipartite. – AveragePythonEnjoyer Jul 21 '22 at 13:46