Affinity Propagation
Message-passing clustering that automatically determines the number of clusters
Affinity Propagation identifies cluster exemplars by passing "responsibility" and "availability" messages between data points. It automatically selects both the number of clusters and the representative exemplar for each cluster.
When to use:
- When the optimal number of clusters is unknown
- Finding representative exemplars is as important as the cluster assignments
- Small-to-medium datasets where O(n²) computation is acceptable
Input: Tabular data with the feature columns defined during training Output: Cluster label for each row (cluster count is determined automatically)
Model Settings (set during training, used at inference)
Damping (default: 0.5, range: 0.5–1.0) Dampens message updates to avoid oscillations. Higher values slow convergence but improve stability.
Max Iter (default: 200) Maximum message-passing iterations.
Convergence Iter (default: 15) Number of iterations with no change before declaring convergence.
Inference Settings
No dedicated inference-time settings. New points are assigned to the nearest trained exemplar.