Clustering Skill

This skill groups data points into clusters based on similarity. Use this skill when users want to segment their data, identify customer groups, find similar patterns, or discover natural groupings in their analytics data.

How It Works

The Clustering skill uses machine learning algorithms to group data points based on similarity across multiple dimensions. When activated, the assistant:

  1. Uses the Visualization skill to prepare the data
  2. Executes clustering algorithms to identify groups
  3. Assigns data points to clusters based on similarity
  4. Creates visualizations showing the clusters
  5. With data sharing enabled, provides meaningful segment labels and summaries

The skill helps you discover natural groupings in your data that might not be immediately obvious, enabling targeted analysis and action for each segment.

Examples

Customer Segmentation

The user asks: Group our customers into segments. The AI Assistant then:

  • runs clustering on customer-level data,
  • creates a visualization of the resulting segments, and
  • labels the segments to describe what makes them different.

Product Clustering

The user asks: Find similar products based on sales patterns. The AI Assistant then:

  • clusters products using their sales behavior,
  • visualizes the product groups, and
  • summarizes the key characteristics of each cluster.

Regional Analysis

The user asks: What are the main clusters in our regional performance? The AI Assistant then:

  • groups regions by similarity in performance metrics,
  • shows the resulting clusters, and
  • highlights what distinguishes each group.

More Example Prompts

  • Find similar customer profiles.
  • Group products by similarity.
  • Find groups of similar regions.
  • Cluster our stores by performance patterns.

Limitations

  • Requires the Visualization skill as a dependency
  • Clustering works best with multiple dimensions and sufficient data points
  • Results interpretation requires data sharing to be enabled
  • Cluster quality depends on the data distribution and chosen metrics

This is an experimental feature that is still under active development. Its behavior may change in future releases.