Clustering - Detects anomalies in batches
Clustering unsupervised learning algorithm for anomaly detection that works on the principle of ‘isolating anomalies”. The algorithm looks at the spread of the dots, and starts to randomly partition them. If the dots are close to each other, it will take the algorithm longer to separate each dot. If the dots are far from each other, less separation is required, and the algorithm works faster. The figure below demonstrates (in a 2D domain) that isolated anomalies typically require less amount of splits.