Converting nested dictionary with overlapping keys to NetworkX

Question

So I have a real mess here, as far as I can see. My goal is a directed and weighted network. The main issue I'm facing is that my output that I'm trying to convert to weighted edges is coming from a 'for' loop, and I tend to get either the first or last output of the loop no matter what I do.

It looks something like this:

import networkx as nx
import matplotlib.pyplot as plt
G2 = nx.DiGraph()
compiledDict = {'A': ['A_1', 'A_2', 'A_3', 'A_4', 'A_5', 'A_6'], 'B': ['B_1', 'B_2', 'B_3', 'B_2', 'B_4', 'B_5', 'B_1', 'B_6', 'B_4']

for x in compiledDict:
    dNotFinalAfterAll={}
    forthelasttime=compiledDict.get(x)
    for item in forthelasttime:
        dNotFinalAfterAll[x]={item:forthelasttime.count(item)}
        print(dNotFinalAfterAll)

And a shortened ver of the output is as follows, each line is one output of the for loop. The goal is to have it where {A:{B:C}} is that A is an outgoing node, B is incoming, and C is the weight:

{'B': {'B_1': 2}}
{'B': {'B_2': 2}}
{'B': {'B_3': 1}}
{'B': {'B_2': 2}}

If there's a way to make it so the duplicates like {'B': {'B_2': 2}} only occur once, that would be good, but as long as it works it can be as ugly as necessary. Needed for proof of concept rather than ongoing process. I want to take this and make it a dataframe. It is not happening. The closest I get is putting this in various places:

G2 = nx.DiGraph((k, v, {'weight': weight}) for k, vs in dNotFinalAfterAll.items() for v, weight in vs.items())
for s, t, w in G2.edges(data=True):
    print(s,t,w)
    G2.add_edge(s,t,weight = w)

Which typically outputs this, whether or not I place it in the for loop:

Z Z_3 {'weight': 1}

Answer 1

Assuming you want to build a weighted DiGraph from your input dictionary, you could take advantage of collections.Counter:

from collections import Counter
import networkx as nx

compiledDict = {'A': ['A_1', 'A_2', 'A_3', 'A_4', 'A_5', 'A_6'],
                'B': ['B_1', 'B_2', 'B_3', 'B_2', 'B_4', 'B_5', 'B_1', 'B_6', 'B_4']}

G2 = nx.DiGraph()

for k1, l in compiledDict.items():
    for k2, w in Counter(l).items():
        G2.add_edge(k1, k2, weight=w)

Or, without the intermediate dictionary:

import networkx as nx

compiledDict = {'A': ['A_1', 'A_2', 'A_3', 'A_4', 'A_5', 'A_6'],
                'B': ['B_1', 'B_2', 'B_3', 'B_2', 'B_4', 'B_5', 'B_1', 'B_6', 'B_4']}

G2 = nx.DiGraph()

for k1, l in compiledDict.items():
    for k2 in l:
        if  G2.has_edge(k1, k2):
            G2[k1][k2]['weight'] += 1
        else:
            G2.add_edge(k1, k2, weight=1)

Graph with weights as edge labels: