Create a Prediction Job for Your Model in Salesforce Data Cloud's Einstein Studio

In my previous posts, I demonstrated how to use Einstein Studio’s No Code Model Builder to train one of the three out-of-the-box regression models available with Salesforce Data Cloud. I also showed how to analyze the training metrics and evaluate the accuracy of your model’s predictions.

Now, it's time to put your regression model to work! The final step is using your model to make predictions with your data. Once you’ve selected your training data, chosen your model, trained it, and assessed its accuracy, you’re ready to feed your model new data to generate predictions. In this post, I’ll guide you through the process in Einstein Studio.

In Einstein Studio you will first need to select your model then navigate to the Integrations sub-tab. Next, you will need to select the button + New Prediction Job.

?

Setting Up Your Prediction Job?

You will then need to select the data model object that contains the data that you will want to make predictions for. This should not be the same set of data that you used to train your predictive model.

?

?

Next you will need to map your data model object’s fields to the variables of your model.

?

?

You will then need to choose how your predictions will be populated to your prediction data model object. You currently have two options to choose between, streaming and batch. Streaming will capture a change event on an input field and output the prediction in near real-time. Batch will only output predictions when you manually initiate an update. If you want predictions to be constantly created as your data is changing or as new data is being populated, you will want to use streaming. If you only want predictions to happen on-demand or when you choose for them to happen you will want to choose batch.?

For a batch job, you can always view your prediction job’s process records by selecting the drop down next to the prediction job and choosing View Last Run.

?

You can also view the prediction for your job using either the Data Explorer or Query Editor to search for the data model object that was created as a result of your prediction job. The data model object will create a new primary key for your record. It will also list the primary key of the original record that was input into the model and the data source. Finally, you will see the prediction depicted as a numerical probability. In this case there is 98.58% that this particular customer will churn. Yikes!

?

?

Conclusion

By following these steps, you can harness the power of Einstein Studio’s No Code Model Builder to make accurate predictions using your Salesforce Data Cloud data. After training your regression model, evaluating its accuracy, and setting up a prediction job, you are fully equipped to leverage insights from your data. Whether you choose real-time predictions with streaming or on-demand batch predictions, the flexibility and usability of Einstein Studio ensure that your business can make data-driven decisions effortlessly.

This predictive capability isn’t just about crunching numbers—it’s about creating actionable insights. Whether you’re forecasting customer churn, identifying sales opportunities, or optimizing business processes, Einstein Studio empowers you to predict outcomes with confidence. Dive in, experiment, and unlock new possibilities for your organization!

Please Subscribe!

If you find this newsletter post or any of my other newsletter posts helpful I would appreciate it if you could subscribe to this newsletter. It's free and only take a few seconds. Thank you very much! If you are already subscribed, thank you for being a subscriber!