projects/cluster-carac

Uncover the unique profile of every data segment

ClusterCarac is an R package that helps you understand what makes each group in your data unique. Whether your groups come from clustering (k-means, hierarchical, etc.) or from predefined categories like customer segments or regions, ClusterCarac reveals the traits that set one group apart from another.

R Characterization EDA Segmentation FactoClass

Turn raw data into clear, interpretable group profiles

Instead of manually inspecting cross-tabs or reading long tables, ClusterCarac answers questions like:

  • “What characterizes Group A, B and C?”
  • “How do high-churn customers differ from low-churn ones?”
  • “What traits truly distinguish this segment?”

Useful in any field where groups appear in your data

  • Customer and market segmentation
  • Social and demographic research
  • Public health and education projects
  • User and product analytics
  • Any dataset with categories or segments that need explanation

In one sentence

ClusterCarac turns groups into stories—showing what defines them, what sets them apart, and what truly matters.

Why I created ClusterCarac

ClusterCarac exists because I wanted to bring back—in a modern, focused way—the spirit of the characterization algorithms originally implemented in FactoClass.

FactoClass, developed at the Universidad Nacional de Colombia with Professor Campo Elías Pardo and collaborators, offered powerful tools for describing groups produced by clustering procedures. Many users asked how to interpret or extract the characterization results, which highlighted a need.

ClusterCarac is my response: not a replacement for FactoClass, but a retro-inspired return to its core ideas—a simplified, refined, and more user-friendly tool dedicated entirely to group characterization.