Research on Lightweight Ensemble Algorithm for Anomaly Detection

As business rapidly evolves, the types of business services become increasingly complex. The conventional approach of manually configuring rules for maintenance and quality assurance systems is gradually proving inadequate, and there is a growing demand for intelligent operations (AIOps) and quality assurance. This paper proposes an anomaly detection algorithm by ensemble various algorithms. Originating from operation and maintenance scenarios, this algorithm is gradually applied to business scenarios. The process involves streaming data collection, data preprocessing, anomaly detection models, and monitoring alerts, thereby realizing multidimensional anomaly detection in both operational and business scenarios. In comparison to traditional monitoring platforms, the proposed algorithm saves manpower and exhibits higher accuracy. Furthermore, when compared to traditional machine learning algorithms, it saves resources and provides faster response times.