Table of Contents

Article

Creating and implementing a decision strategy for predictive analytics

To apply the predictive power of a model to predict customer behavior, create a strategy that references that model.

Use case

uPlusTelco wants to improve the experience of their customer support by predicting the reason for each customer call. To achieve that goal, the data analytics team built a predictive model and uploaded the model to Prediction Studio.

Your task as a system architect is to create a decision strategy that uses that model and to deploy the strategy in a decision data flow.

Before you begin

Import a PMML model and map its predictors to Pega Platform™ properties. See Importing and configuring a PMML model.

Creating a decision strategy with a predictive model

  1. In Dev Studio, click Create > Decision > Strategy.
  2. In the New tab, configure the basic strategy settings:
    1. In the Label field, enter a name, for example:

      Monitor My Call Context

    2. In the Context section, enter the same class and open ruleset version as for the model that you imported, for example:

      Apply to: DMOrg-DMSample-Data-Customer

      Development branch: No branch

      Add to ruleset: DMSample

      Ruleset version: 08-01-99

      Thumbnail
      Creating a strategy - basic settings
  3. Verify the initial settings, and then click Create and open.
  4. On the strategy canvas, reference the predictive model that you want to monitor:
    1. Add the Predictive Model shape by right-clicking the canvas and selecting Decision Analytics > Predictive Model.
    2. Connect the Predictive Model shape to the Results shape.
    3. Double-click the Predictive Model shape.
    4. In the Predictive model properties dialog box, enter a name for the model, for example:

      Predict My Call Context

    5. In the Predictive model field, select the predictive model that you want to use for predictions, for example:

      PredictMyCallContext

    6. Verify the settings and click Submit.
      Thumbnail
      Predictive model properties
  5. On the strategy canvas, verify that you have all the following shapes and click Save:
    Thumbnail
    Sample decision strategy

Deploying a strategy in a decision data flow

Add your strategy in a decision data flow to make predictions with the predictive model that you created.

For the purpose of this tutorial, use an existing data flow - Monitor Predictive Models - as the baseline. This model provides you with preconfigured elements that you can use to effectively run your strategy:

  • The source data flow that gathers customer data for your strategy.
  • The target data set for collecting decision results. The same data set is later used as the source in the response strategy - Set Responses to Monitor Models.

After running this decision data flow with your own strategy, the Decision Results data set contains a list of interactions with individual pxInteractionID property values for later analysis in predictive model monitoring.

  1. In Dev Studio, click Records > Data Model > Data Flow.
  2. On the list of the Data Flow rule instances, click MonitorPredictiveModels.
  3. On the data flow canvas, click Save as.
  4. On the New tab, configure the basic data flow settings:
    1. In the Label field, enter a name, for example:

      Monitor My Predictive Model

    2. In the Context section, enter the same class and open ruleset version as for the strategy that you created, for example:

      Apply to: DMOrg-DMSample-Data-Customer

      Development branch: No branch

      Add to ruleset: DMSample

      Ruleset version: 08-01-99

  5. Verify the initial settings, and then click Create and open.

  6. On the data flow canvas, reference your decision strategy:

    1. Double-click the strategy shape.

    2. In the Edit: Decision strategy configurations dialog box, in the Strategy field, delete the current entry and enter the name of your decision strategy, for example:

      MonitorMyCallContext

    3. In the Mode field, select Make decision and store data for later response capture and enter how long you want to store adaptive inputs and strategy results, for example:

      1 day

      By turning this setting on, you configure the strategy to store the decision data in pxDecisionResults for delayed adaptive model learning and for issuing a response capture at a later time.

    4. Leave all other settings unchanged. Click Submit.

      Thumbnail
      Editing a decision data flow - strategy configuration

       

  7. On the strategy canvas, verify that you have all the following shapes and click Save:

Thumbnail
Sample decision data flow

Conclusions

You have created a strategy that uses the Predict My Call Context model to predict future customer behavior, and you have referenced that strategy in a decision data flow.

What do to next

Define the model objective and outcome to monitor by creating a response strategy. See Creating a response strategy.

To view the main process outline for this tutorial, see Monitoring predictive models.

Published December 5, 2018 — Updated March 22, 2019

Related Content

Have a question? Get answers now.

Visit the Pega Support Community to ask questions, engage in discussions, and help others.