The early detection of anomalies is vital for ensuring undisrupted business and efficient troubleshooting, and anomaly detection has many applications, such as tracking business KPIs or fraud spotting in credit card transactions. Unfortunately, there’s no one best way to detect anomalies across a variety of domains. Such a methodology is a myth given that time series can display a wide range of behaviors. In addition, what behavior is anomalous can differ from application to application.
Jonathan Merriman and Cynthia Freeman introduce a framework to determine the best anomaly detection method for the application based on time series characteristics (e.g., seasonality, concept drift, etc.). It allows you to plug in detection methods as well as how you want to evaluate them. This framework is not limited to any particular anomaly detection or evaluation methods. In addition, the framework can be applied to a broad set of time series classes.
Jon Merriman is a senior software engineer and researcher at Verint Intelligent Self-Service, where he works on core natural language understanding capabilities for dialogue systems. His primary focus is on algorithms and machine learning theory for text and speech analysis.
Cynthia Freeman is a research engineer at Verint Intelligent Self-Service, a developer of conversational AI systems. She holds an MS in applied mathematics from the University of Washington and a BS in mathematics from Gonzaga University and is currently pursuing her PhD in computer science at the University of New Mexico, where she works on time series analysis and developing new anomaly detection methods.
©2019, O'Reilly Media, Inc. • (800) 889-8969 or (707) 827-7019 • Monday-Friday 7:30am-5pm PT • All trademarks and registered trademarks appearing on oreilly.com are the property of their respective owners. • firstname.lastname@example.org