Enables event stream processing and on-line scoring based on analytical models built upon data coming from the stream of events.
Processing the events (in JSON format) sent from your application via HTTP (REST API). Support for SQL language (including analytical functions).
Automatic building of predictive models based on streaming data – fully integrated with our Automatic Business Modeler (ABM).
Building an on-line user profile (list of aggregates) easily accessible via the REST API.
The ON-LINE mode
- The values of all aggregates used by an active and deployed models are updated based on each incoming event
- Updated aggregates with predefined parameters are then used to trigger an on-line scoring for models that meet required conditions
- For each incoming event a response (JSON) containing the score values for every triggered model is returned
- Based on the returned scores, the application can take an appropriate action
The OFF-LINE mode
- Every event is saved in an event repository
- Based on the saved events the offline-process calculates aggregates which are used to build predictive models
- The models are automatically deployed and can be used for on-line scoring
- The offline processing can be scheduled to run at the specified periods of time