Anomaly Detection in Streaming Data using Isolation Forest
Mohammed Shaker Kareem, Lamia AbedNoor Muhammed · 2024
In the era of constant data generation and exchange, the concept of data streams, which entails the rapid and organized transmission of vast volumes of data online, has become prevalent. Data streams are susceptible to concept drift, where data distribution changes over time, and are often high-dimensional. Anomaly detection in data streams is a critical task with applications in diverse domains such as fraud detection, cyber security, fault monitoring, and military surveillance. Isolation Forest (iForest) has gained popularity as an anomaly detection method due to its ability to effectively capture rare and distinct anomalies, making minimal assumptions about data properties, and offering linear time complexity.