Skip to main content

AI Builder now supports numerical prediction (Preview)

Headshot of article author Norah Liu


AI Builder prediction models now support a preview capability to predict a number. Now you can use AI builder intelligence to predict things like product ratings, price estimates, time to completion, and so on.

In this post, we demonstrate how numerical prediction works by building an example of an end to end scenario where we’d use AI Builder to help an online business to optimize purchase conversion rate.


Let’s say our online business is quite competitive, and there’s a need to improve purchase conversion. How can we increase the percentage of customers visiting our website who then make a purchase?

One approach is to identify customers visiting a product page that have a high probability to exit the page before completing a purchase. Using AI builder prediction modeling, we can identify this customer as ‘likely to drop’, and then trigger a Power Automate flow to send a coupon to this customer. This targeted interaction enabled by AI Builder prediction modeling can help convert site visitors to customers.

Identify target field

For this scenario, we’ll use the  Online Shopper Intention entity. This entity contains historical online shopper behavioral data from the past year. There are two fields which are particularly related to the issue we want to predict.

ExitRates: Probability that a user would leave the current webpage.

BounceRates: Probability that a user would navigate away from the current website after viewing only one page

Let’s use ExitRates, as users who viewed more pages before they exit are probably better candidates to convert to customers.

Create prediction model

First, let’s create a new prediction model from AI Builder section of Power Apps. More information about how to do this: Creating a prediction model.

We use my Online Shopper Intention entity and the ExitRates field. Note, numerical prediction is still in preview, so numerical fields are annotated with a ‘Glimmer’ in the Field drop down menu.

Next, we’ll exclude BounceRates from the training fields as it might have a high correlation with exit rates.

Here, let’s skip adding a filter as this data should be sufficient to train the model.

Once all that is done, it is time to train the model. For this model, we have a linear performance score of 83. Linear performance scores measure closeness between predicted data and actual data. It can be between 0 – 100%, with higher values indicating the predicted value is closer to the real value.  Generally, a higher score means the model should perform better. However, be wary of perfect scores, as this can indicate an overfit model.

Publish and use your model

As part of publishing, we actually create three outputs:

  • Predicted – The predicted outcome
  • Probability – The probability of the true outcome, from 0 – 1. But this field would be empty as it doesn’t apply when you’re predicting a number.
  • Explanation – A JSON text field that contains the record level influencers

After model is published, select Use model, and then select Run now to get the prediction results immediately. Let’s give it a few minutes to let the model run on our data set.  The model will predict for any records with empty ExitRates field.

After above step is done, we can use the view feature to quickly see the predicted output. Refer to this document for more info.

Now, we’re ready to connect the prediction model to our Power Automate flow. For this example, we can set the flow to be triggered when the ExitRates is higher than 15%, and send an email to this customer with coupon information.

Now, go ahead and try this scenario on your own!  Your feedback will  help us continue to build on and improve AI Builder. We want to hear from you!

You can refer to this document for guidance about how to import the sample data set so you can try it on your own!