Data preparation

Before you create your prediction model, you'll want to make sure your data is in Microsoft Dataverse and that it's in the correct format.

Create your custom table

Do you have data that you want to import into Dataverse for training in AI Builder? First, you have to create a table. In this example, we'll provide a solution that has predefined custom tables. To use your own data, create a custom table and use it instead of the example used here.


For best results, use a dataset that is less than 1.5 GB in size. Otherwise, AI Builder will use only 1.5 GB of your data to train and predict. Because you can’t control which data that exceeds the 1.5 GB limit won't be used, you should optimize your data to stay under 1.5 GB.

Sample dataset for prediction model

  1. Download the AI Builder sample dataset package:

    1. Select
    2. Select the Download button.

    Screenshot of the download screen in GitHub.

  2. Ensure that the file isn't blocked after you download. To do this:

    1. In the Downloads folder, find the downloaded zip file, right-click, and then select Properties.
    2. On the General tab, select the Unblock checkbox, and then select Apply.

    Screenshot of the .zip file Properties.

  3. Extract the .zip file, and look for PackageDeployer.exe in the extracted folder.

    Screenshot showing PackageDeployer.exe selected.

  4. Run PackageDeployer.exe. The following screen will appear.

    Screenshot of the Package Deployer landing page.

  5. Select Continue.

  6. Select Office 365, and then select Login.

    Screenshot of signing in to the Package Deployer.

  7. Enter the credentials that you use to sign in to the Power Apps maker portal, and then select Next.

    Screenshot of entering your credentials.

  8. If the sign-in is successful, you'll see the Welcome screen. Read the message, and then select Next.

    Screenshot of installing the AI Builder prediction sample dataset.

  9. On the Ready to Install screen, make sure you're installing the solution in the correct environment and then select Next.

    Screenshot showing the solution package to be installed and the organization name.

  10. On the Reading AIB Prediction Dataset Installer Configuration screen, read the summary information for the data and solutions being imported and then select Next.

    It will take a few minutes to import the data. As each step is completed successfully, you'll see a green circle with a check mark next to the step.

    Screenshot of the summary of the AI Builder prediction dataset installer configuration.

  11. On the Installation Complete screen, select Finish.

    Screenshot showing installation complete.

How you can use the solutions you installed

The sample dataset installs two solutions in your environment. It also installs sample data for the included entities:

  • Brazilian Commerce: Use for predicting multiple outcomes. Select BC Orders as the table, and Delivery Timeliness as the column when you pick what you want to predict.

  • AI Builder Online Shopper Intention: Use for binary prediction and numerical prediction.

    • Select Online Shopper Intentions as the table, and Revenue (Label) as the column if you want to try out binary prediction.
    • Select Online Shopper Intentions as the table. Also select ExitRates or BounceRates as the column if you want to try out numerical prediction.

If you need help with creating a prediction model, follow the instructions in Create a prediction model.

Screenshot of solution details.

You're now ready to go to the next step.

Next step

Create a prediction model