constructing subgraphs with thicker edges (networkX)-CodePudding

merged_table=

ID         0   1   2   3   4   5   6   7   8   9  10  11  12
131X1319   1  14   6  16   1  10   8  15  15  17  15  18  16
13212YX3   1   1   4   8  11   9  14   7   0   3   0  17  13
13216131   1   1  13   9  15  17   0   9   3  15  11   8  10
132921W6   1  14  10   4  18   7   8  15  15  17  15  18  16
132W22YY   0  13  12   9  15   5   0  13   3  15  11  15   6
132X219Y  16   5   8  11  15   3   6   2  13   6   4  14  10
132Y231B  17  18  17  17  19  18  20  15  15  17  15  18  16

I have a data frame like this and I have constructed a graph, in which the edges get thicker if two nodes occur more than once. This is the graph I have generated:

columns=list(merged_table.columns.values)
ct = Counter(p for i in range(len(columns)-2) 
               for p in zip(merged_table[columns[i]],merged_table[columns[i 1]]))
pattern_g = nx.DiGraph()
pattern_g.add_edges_from(ct)
width = [ct[p] for p in pattern_g.edges] 
nx.draw(pattern_g, node_color = 'orange', with_labels=True, width = width)
plt.show()

Now, I want to draw subgraphs of nodes that have a thicker edge in the graph. if the flow of the nodes is more than two it's better. I tried using strongly_connected_componenets but it is printing all the nodes and edges of the graph.

CodePudding user response：

I assume you have in mind some kind of side-by-side comparison. With that in mind, here's a modification of your script that produces drawings of both graphs with the same layout.

columns=list(merged_table.columns.values)
ct = Counter(p for i in range(len(columns)-2) 
               for p in zip(merged_table[columns[i]],merged_table[columns[i 1]]))
pattern_g = nx.DiGraph()
pattern_g.add_edges_from(ct)

# get layout for use in both graph drawings
pos = nx.spring_layout(pattern_g)

# Draw first graph, using frequency of each pair as edge-width
width = [ct[p] for p in pattern_g.edges] 
nx.draw(pattern_g, pos = pos, node_color = 'orange', with_labels=True, width = width)
plt.show()

# Build and draw second graph, excluding edges with wgts below threshold 'thr'
thr = 1
pattern_subg = nx.DiGraph()
for e,wgt in ct.items():
    if wgt>thr:
        pattern_subg.add_edge(*e)
width = [ct[p] for p in pattern_subg.edges]
nx.draw(pattern_subg, pos = pos, node_color = 'orange', with_labels=True, width = width)
plt.show()