P97Unsupervised LearningAdvancedDiagnose18 min

Diagnose Cluster Instability After Adding Encoded Features

Diagnose instability caused by representation and distance choices in a clustering pipeline.

#diagnose#clustering#encoded features

Scenario

Customer clustering was stable with spend and visit frequency. After adding 40 one-hot product-category columns, 35% of assignments change across seeds, cluster sizes swing from 8% to 44%, and encoded dimensions account for most pairwise Euclidean distance.

Problem statement

Diagnose plausible causes and propose checks before presenting the clusters as segments.

Your Task

  1. Assess scaling, representation, and dimensionality.
  2. Consider initialization and weak structure.
  3. Recommend stability-focused validation.

Related Concepts

ClusteringFeature scalingStability analysis

Ready to count this problem?

Mark it complete when you have reasoned through the solution in your own words.