Time Series Anomaly Detection

Detect abnormal data behavior in time series using tools such as matrix profiling algorithms and specialized deep learning, machine learning, and statistical models that identify deviations from normal behavior

Anomaly detection in time series is the process of identifying anomalous data patterns by thoroughly characterizing normal behavior and detecting deviations from that behavior. The anomaly detection techniques here represent two major approaches for detecting anomalies:

  1. Time Series Anomaly Detector Models—These models draw from machine learning, deep learning, and statistical algorithms, and apply them specifically to subsequence-level anomalies that result in anomalous patterns across multiple points within a time series. This approach provides the advantage of requiring only normal data for system characterization, with just a relatively small amount of anomalous data for testing. These models, and associated functions and app, are provided in the Time Series Anomaly Detection for MATLAB® support package.

  2. Distance Profiling— These methods base anomaly detection on pattern matching at the subsequence level. These functions are provided in the base Predictive Maintenance Toolbox™ toolbox.

Both approaches perform detection at both he subsequence level in a time series and the point level.

Categories