Automated Anomaly Detection with Medoid Clustering in n8n

This n8n workflow is designed to enhance your anomaly detection capabilities by intelligently identifying 'medoids' – representative data points within clusters. Leveraging Qdrant for efficient vector search, the workflow first establishes cluster centroids and then refines these into medoids. It calculates a distance matrix to find the most central point within each cluster and marks it as a medoid in your Qdrant collection. Subsequently, it retrieves the vector of this medoid and performs a targeted search to establish an anomaly threshold, effectively distinguishing normal data from outliers. This approach is particularly useful for datasets like crop information, where identifying deviations from typical patterns is critical for monitoring and management.