Complex Network Analysis in Python

Name the Components

The new induced graph nicely reflects the macroscopic structure of the original product network. It has only eighteen nodes and twenty-nine edges. The nodes are nameless so far, and we need to give them names. Having no better source of labels than the product categories, we select the most popular category within each induced node as the node label. We need an auxiliary function to obtain the name of the dominant category in a community. The Sephora website reports category names as colon-separated hierarchical paths. To save space in the future printout, we keep only the last path component:

products.py

def top_cat_label(community_subgraph):
    items = [atts["category"] for _, atts
             in community_subgraph.nodes(data=True)] ...