Alarm Rule

Metatron Anomaly makes it easy for users to easily create and manage rules that trigger alarms in their time-series data. Metatron Anomaly’s alarm rules have the following characteristics:

  • Machine learning based on unsupervised learning for all real-time data without error history

  • Easy and fast alarm rule creation in 3-step

  • 통계 기반의 single 단변량 분석 알고리즘 7종 및 다변량 분석 알고리즘 1종의 기본 예측 모델 제공

  • Automatic model learning and optimal model recommendation

  • Supports model re-learning when applied model accuracy decreases

The structure of this unit is as follows.