Data Flow
Each component in the API was designed to be pluggable allowing users to mix and match between the framework's built-in components and user provided custom components
-
Events: Input Dataset each record has a timestamp
-
Windowing: Events are grouped into windows for processing e.g. 1 hour, 24 hours
-
Aggregation: Data structure and processing of the grouped events collected in each window
-
Anomaly Detection Model: Analyzes an aggregation and decides if it’s an anomaly, produces reports
-
Reports: Output Dataset of an anomaly detection analysis, produced when examining each aggregation