Response capture works differently after upgrade from Pega Platform version 7.x or 8.1 to a later version
Before the introduction of model execution results, models were updated with responses even on a subset of context keys.
For example, before Pega Platform version 8.2, for an adaptive model rule with three context keys: pyIssue, pyGroup, and pyName, three models were created with the following properties:
- pyIssue, pyGroup
- pyGroup and pyName
If a response was sent with the pyIssue and pyGroup properties, then all three models received the response.
This behavior changed with the introduction of model execution results for response handling for adaptive models. In Pega Platform versions 8.2 and later, in a similar scenario, only the model with the pyIssue and pyGroup properties receives the response.
With model execution results, it has become easier for a single response to update all component models. When models are executed in the course of a strategy result, the model execution results are passed along and when they are part of the outputs, they are stored in pxDecisionResults.
When a response is captured, all the models that have executions related to the strategy result are updated (pyOutcome should still match, and optionally pyPrediction).
If you want multiple models to learn from the same response, perform the following actions:
In a Data Join component in your decision strategy, select Exclude model results from strategy_shape_name if you only need the model results from the source components.For more information, see Enrichment.
In a Group By component, specify which model results to include by selecting Include model results for or Include all model results in group.For more information, see Aggregation.