How to making edges from dataframe in NX-CodePudding

I have a dataframe that contains movies, actors names etc. And it has 41k rows.

I'm planning to make a graph from NX library and I want to use actors as nodes, and make edges if they are played in a same movie. I tried to make it dataframe and do it with for loops but I couldn't. Can you help me?

Edit: I want to make a graph like this:

CodePudding user response：

IIUC, let's try something like this using networkx and itertools libraries:

from itertools import tee
import networkx as nx
import pandas as pd

df = pd.DataFrame({'Movie': [*'AAABBCCCDD'],
                  'Actor':[1,2,3,2,5,7,8,9,10,8]})

def pairwise(iterable):
    "s -> (s0,s1), (s1,s2), (s2, s3), ..."
    a, b = tee(iterable)
    next(b, None)
    return zip(a, b)


G = nx.Graph()
for _, s in df.groupby('Movie'):
    if s.shape[0] > 1:
        [G.add_edge(*i) for i in pairwise(s['Actor'])]
    else:
        G.add_node(s['Actor'].iloc[0])
        
nx.draw_networkx(G)

[list(i) for i in nx.connected_components(G)]

Output:

And, actor groups:

[[1, 2, 3, 5], [8, 9, 10, 7]]