How Far Should We Look Back to Achieve Effective Real-Time Time-Series Anomaly Detection?

Anomaly detection is the process of identifying unexpected events or ab-normalities in data, and it has been applied many different areas such as system monitoring, fraud detection, healthcare, intrusion etc. Providing real-time, lightweight, proactive anomaly for time series with neither human intervention nor domain knowledge could be highly valuable since reduces effort enables appropriate countermeasures to undertaken before a disastrous event occurs. To our knowledge, RePAD (Real-time Proactive Detection algorithm) generic approach all above-mentioned features. achieve real-time lightweight utilizes Long Short-Term Memory (LSTM) detect whether not each upcoming data point anomalous based on short-term historical points. However, unclear that how amounts points affect performance RePAD. Therefore, this paper, we investigate impact by introducing set metrics cover novel accuracy measures, efficiency, readiness, resource consumption, Empirical experiments real-world datasets are conducted evaluate scenarios, experimental results presented discussed.