@RajeshP16753019

To add more Interaction History (IH) based predictors to your adaptive models here are some recommended best practices: 1. Use IH Summaries: Utilize IH summaries to preprocess and summarize IH data for predictions. This helps in extracting fields such as the number of recent purchases the time since the last purchase and so on. 2. Enable Predefined IH Predictors: Several predictors based on IH summaries are enabled by default for all new adaptive models. These include fields like the group referenced in the last interaction and the number of days since the last interaction. 3. Materialized Data Sets: Use materialized data sets for IH summaries as they are preaggregated and always available for use which saves processing time. Non materialized data sets are also available but require more time to process as they query the database for the data in use. 4. Analyze Additional Predictors: Review the additional set of predictors based on a predefined IH summary. These predictors start with the 'IH.' prefix in the Adaptive Model monitoring reports. Analyze the possible values for fields like Channel Direction and Outcome in the Interaction History to understand why the number of possible combinations is high. 5. Add Uncorrelated Predictors: It is highly recommended to add many uncorrelated predictors as the models will figure out which ones to use. This can improve the predictive power of your adaptive models. 6. Parameterized Predictors: Consider creating parameterized predictors to improve the predictive power of your adaptive models. These are input fields that are not directly available as properties in the customer data model but can be made available to adaptive models by defining them as parameterized predictors.

⚠ This is a GenAI-powered tool. All generated answers require validation against the provided references.

Enabling predefined Interaction History predictors for existing adaptive models

Adding adaptive model predictors > Predictor types based on data sources

Best practices for choosing predictors

Configuring pyIHSummary for use as a predictor in an adaptive model

Predictors and outcomes of an adaptive model