Create/Edit Machine Learning Strategy | Vantage CX - Creating or Editing a Machine Learning Strategy - Vantage Customer Experience

Vantage Customer Experience User Guide

Deployment
VantageCloud
VantageCore
Edition
Enterprise
IntelliFlex
VMware
Product
Vantage Customer Experience
Release Number
1.6
Published
October 2023
Language
English (United States)
Last Update
2023-10-26
dita:mapPath
hbt1563223944614.ditamap
dita:ditavalPath
oyf1565965838286.ditaval
dita:id
tmo1562969305444
Product Category
Teradata Applications

After defining a strategy, you can include it in arbitration formulas.

You must promote the ORG to Production to update the cache and make the new or changed strategy available. See Promoting an Environment.

  1. Select "" > Machine Learning.
  2. Create a new strategy or edit an existing one:
    • Select Create New.
    • For an existing strategy, select Edit.
  3. Enter (or edit) the value for each field, then select Next.
    All fields are required except Advanced Options. If you're editing an existing strategy, you can skip to the fields you are modifying.
    Field Description
    Name your strategy Enter a unique name for this strategy.
    What responses should this strategy target? Select one or more of the system-defined responses. The strategy will generate models used to score messages (prediction targets) according to their likelihood to yield these selected responses.
    What do you want to predict? Select a target to score. The selected attributes will be scored as part of the arbitration step within the decision process. You can select:
    • All messages or only required messages
    • All message groups or only required message groups
    • Personalization Attribute of lookup type

      Any message with predefined attribute values. Qualifying messages with the selected attribute-value combinations are arbitrated according to the score of the combination.

    What factors influence your strategy? Selection attributes that should be considered when generating a machine learning model.

    Within real-time decisioning, models are used to generate the target arbitration score based on the values of the selected attributes.

    [Optional] Review your strategy > Advanced Options Select Advanced options to modify the default threshold to trigger building the machine learning model.
    • For previously generated models: Change percent to trigger model rebuild

      Percentage is based on the total number of new interactions that would be captured in the prospective model. Default: 50.

    • For models that have never been generated: Minimum number of responses

      Minimum number of accept responses. This ensures that a significant size history is reached for generated messages prior to model building and training. Default: 100.

  4. Verify the strategy content is correct, then select Create This Strategy or Update This Strategy.