Maintaining Objects in Intelligence Connector 

Purpose

This process enables you to identify various objects and tasks that you need to specify in the Intelligence Connector before the application can perform real-time predictions. 

Procedure

Before an application can execute a prediction task or a KPI task using the Application API, you must perform the following actions:

1. Specifying KPI Sets: You must have specified the relevant KPI sets in the CRM application server to be able to execute a KPI task. In the CRM application server, choose Analytics ® Edit KPI Set to specify the KPI sets required by your application. Choose Analytics  ® Edit Key Filed for KPI Set to specify the key fields. For further details, see KPI Sets.

 The CRM application server only contains the metadata of KPI sets and possibly local caches for these KPI sets. The actual KPIs are taken from SAP BW by executing queries or access to an ODS object.

2. Specifying Server Connections. Since the actual predictions are done by data mining servers, to execute a prediction task, you must specify the server connections for the Intelligence Connector to be able to connect to the data mining servers. To execute a KPI task, you must specify the connection to the CRM application server where the KPIs have been specified. Even though multiple connections are supported to the CRM application server is supported, only one connection can be activated at one time. For further details, see Server Types.
3. Creating Applications: To execute a real-time task, you must create the application in the Intelligence Connector. For further information, see Application.
4. Deploy Models: Before a mining model can be used for prediction in a prediction task, you must deploy the model from the source server to the CRM Intelligence Connector. For further details, see Models and Model Deployment and Prediction Processing.
5. Specify Real Time Tasks: You must create a prediction task or a KPI task in the Intelligence Connector. To specify a prediction task, you must deploy the mining model to be used for prediction, specify the data mining server to be used for the processing and map the application fields to the model and prediction output fields. For a KPI task, you must specify the KPI set to be used and map key fields and KPI fields of the KPI set. For further details, see Prediction Task,  KPI Task and Executing Prediction and KPI Task.

Result

You will be able to execute a prediction task or a KPI task from the Intelligence Connector. For the application to execute the real-time task, input data must be collected such as customer age or income.  Although prediction tasks do not require values for all the input fields, the quality of the prediction highly depends on the number of provided input values. To execute a KPI task, all key values must be collected.