COMPEL Glossary / unsupervised-learning
Unsupervised Learning
Unsupervised learning is a machine learning approach that discovers hidden patterns and structures in data without pre-labeled examples.
What this means in practice
Unlike supervised learning, there is no 'correct answer' to learn from -- the model finds groupings, anomalies, and relationships on its own. Common techniques include clustering (grouping similar data points), dimensionality reduction (simplifying complex data), and anomaly detection (finding unusual patterns). Unsupervised learning is most valuable when you know patterns exist in your data but do not know what they are -- for example, discovering customer segments, detecting network intrusions, or identifying manufacturing defects. For transformation leaders, unsupervised learning is an exploration tool that often generates insights which then inform supervised learning projects or human decision-making.
Why it matters
Unsupervised learning discovers hidden patterns organizations did not know existed, making it an invaluable exploration tool for customer segmentation, anomaly detection, and pattern discovery. Unlike supervised learning, it does not require labeled data, removing a major cost and time barrier. However, unsupervised results require human interpretation and domain expertise to translate discovered patterns into actionable business insights.
How COMPEL uses it
Unsupervised learning use cases are evaluated during the Model stage for their exploration and discovery potential. The Calibrate stage assesses data readiness for unsupervised analysis, particularly data quality and volume. Unsupervised insights often generate hypotheses that inform supervised learning projects in subsequent COMPEL cycles. The Evaluate stage measures whether unsupervised discoveries translated into business decisions or informed future use case selection.
Related Terms
Other glossary terms mentioned in this entry's definition and context.