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
- Assess scaling, representation, and dimensionality.
- Consider initialization and weak structure.
- 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.