Anomaly Detection on Univariate Time Series Data Using Exponentially Weighted Moving Average (AnEWMA)

Anomaly detection in time series data is a critical task with wide-ranging applications in industries such as finance, cybersecurity, healthcare, and manufacturing. It involves the identification of data points or patterns that deviate significantly from the expected behavior, thereby ensuring the integrity and reliability of data analysis and decision-making processes. Several methods have been developed to address this challenge, each offering unique advantages and addressing different aspects of the problem, ranging from statistical methods, to machine learning techniques, and dynamic time warping methods. In this work, we present a novel Anomaly Detection approach (AnEWMA) able to identify anomalies through the application of the Exponentially Weighted Moving Average (EWMA). AnEWMA leverages the responsiveness of EWMA to subtle shifts in data trends, enabling the detection of anomalies in a lightweight and computationally efficient manner. AnEWMA adjusts the control limits of the monitoring system using tuned heuristic multipliers. Traditional methods often rely on fixed control limits, which can lead to a high rate of false positives or missed anomalies, especially in the presence of noisy or non-stationary data. The proposed AnEWMA algorithm shows promising results when compared with state-of-the-art unsupervised and semi-supervised anomaly detection methods using stream data from popular Benchmarks.