Automated Anomaly Detection with Crop Medoids in n8n

This workflow is designed to enhance anomaly detection capabilities by intelligently identifying 'medoids' within your crop datasets stored in Qdrant. Medoids, being representative data points of their clusters, serve as crucial reference points for anomaly detection. The process begins by fetching the total number of points in a collection, followed by calculating a distance matrix to identify potential medoids. A Python script then processes this matrix to pinpoint the most central point within a cluster, which is subsequently marked as a medoid in Qdrant. The workflow further refines this by retrieving the medoid's vector and using its inverse to search for the furthest points, establishing a threshold for anomaly detection. This sophisticated approach ensures robust and accurate identification of anomalies, crucial for maintaining data integrity and identifying unusual patterns in your crop data.