Find patterns in data without labels.
Unsupervised Learning Specialist
scikit-learn, UMAP, HDBSCAN
Best for
- ▸Customer segmentation for targeted marketing campaigns using density-based clustering
- ▸Fraud detection using isolation forests and anomaly scoring thresholds
- ▸Market basket analysis through association rule mining and pattern discovery
- ▸Dimensionality reduction for high-dimensional sensor data visualization
What you'll get
- ▸Python code implementing HDBSCAN clustering with parameter selection, cluster validation metrics, and visualization using UMAP embeddings
- ▸Comparative analysis of multiple clustering algorithms with silhouette scores, explained variance ratios, and business-interpretable cluster profiles
- ▸Anomaly detection pipeline with Isolation Forest, contamination parameter tuning, and threshold selection based on business constraints
Clean, preprocessed datasets with appropriate scaling and missing value handling, along with clear business objectives for what patterns or structures to discover.
Cluster assignments, anomaly scores, reduced-dimension embeddings, and diagnostic metrics with recommendations for hyperparameter tuning and algorithm selection.
What's inside
“You are an Unsupervised Learning Specialist. You discover hidden structure in unlabeled data via clustering, anomaly detection, dimensionality reduction, and density estimation. - Apply algorithm-specific geometric assumptions and failure modes: you know K-Means assumes spherical clusters and DBSCAN...”
Covers
Not designed for ↓
- ×Predicting specific target values or classifications (that's supervised learning)
- ×Real-time streaming anomaly detection without batch processing setup
- ×Generating synthetic data or creating new samples from learned patterns
- ×Providing business explanations for why clusters formed (requires domain expertise)
SupaScore
84.88▼
Evidence Policy
Standard: no explicit evidence policy.
Research Foundation: 6 sources (2 official docs, 4 paper)
This skill was developed through independent research and synthesis. SupaSkills is not affiliated with or endorsed by any cited author or organisation.
Version History
v5.5 final distill
Pipeline v4: rebuilt with 3 helper skills
Initial version
Prerequisites
Use these skills first for best results.
Works well with
Need more depth?
Specialist skills that go deeper in areas this skill touches.
Common Workflows
Customer Analytics Pipeline
Complete customer discovery workflow from data preparation through segmentation to marketing strategy
Fraud Detection System
Build and deploy unsupervised fraud detection from exploration to production monitoring
© 2026 Kill The Dragon GmbH. This skill and its system prompt are protected by copyright. Unauthorised redistribution is prohibited. Terms of Service · Legal Notice