AD2S experiments

For Internet-based service companies, anomaly detection on data streams is critical in troubleshooting, seeking to maintain service quality and reliability. Most of known detection methods have an underlying assumption that the data are always continuous. In practical applications, however, we learn that many real-world data are sporadic. It incurs particular challenges for the task of anomaly detection, for which the common preprocessing of downsampling on sporadic data can omit potential anomalies and delay alarms. In this paper, we propose an adaptive anomaly detection method on sporadic data streams named AD2S. It consists of two modules: a monitor module to continuously and adaptively determine the measure windows for observations, and a detection module that utilizes an isolation partition strategy to estimate the anomaly degree of each incoming observation. Our analysis demonstrates that the proposed method has constant amortized time and space complexity. Based on experimental results on both synthetic and public real-world datasets, our method outperforms other state-of-the-art methods in anomaly detection on sporadic data streams, and the code is open-sourced.