A sample representation of Twitter information bubble. Generated by ChatGPT.

analyse_cluster.py

# Analyze the clusters and connections in the graph
for node, cluster in partition.items():
  if cluster == 1:
    # Calculate the percentage of nodes that represent Tweets and the percentage of nodes that represent people you follow
    num_tweets = len([n for n in cluster1 if n['type'] == 'tweet'])
    num_following = len([n for n in cluster1 if n['type'] == 'person'])
    tweet_percentage = num_tweets / len(cluster1)
    following_percentage = num_following / len(cluster1)
    print(f'Cluster 1: {tweet_percentage}% Tweets, {following_percentage}% people you follow')

    # Calculate the percentage of connections that exist within the cluster and the percentage of connections that exist between the cluster and other clusters
    num_intra_connections = len([e for e in G.edges if e[0] in cluster1 and e[1] in cluster1])
    num_inter_connections = len([e for e in G.edges if e[0] in cluster1 and e[1] not in cluster1])
    intra_percentage = num_intra_connections / len(cluster1)
    inter_percentage = num_inter_connections / len(cluster1)
    print(f'Cluster 1: {intra_percentage}% intra-cluster connections, {inter_percentage}% inter-cluster connections')
  elif cluster == 2:
    ...

twitter_cluster_ex.py

import networkx as nx
from sklearn.preprocessing import StandardScaler
from sklearn_extra.cluster import KMedoids

# Collect data about the Tweets you've liked and the people you follow
tweets = ['tweet1', 'tweet2', 'tweet3']
following = ['person1', 'person2', 'person3']

# Preprocess the data
data = [[1, 2, 3], [4, 5, 6], [7, 8, 9]]  # example data
scaler = StandardScaler()
data_scaled = scaler.fit_transform(data)

# Create a graph object
G = nx.Graph()

# Add nodes to the graph
for tweet in tweets:
  G.add_node(tweet, type='tweet')
for person in following:
  G.add_node(person, type='person')

# Add edges to the graph
for tweet in tweets:
  for person in following:
    G.add_edge(tweet, person)

# Use modularity maximization to group the nodes into clusters
partition = community.best_partition(G)

# Visualize the graph using D3.js
d3.render(G, 'graph.html')

# Analyze the clusters and connections in the graph
for node, cluster in partition.items():
  if cluster == 1:
    print(f'{node} belongs to cluster 1')
  elif cluster == 2:
    print(f'{node} belongs to cluster 2')
  ...