Check Your Data

Data Checker automatically validates your data and alerts you as you build a prediction. It provides feedback in real time so you can catch issues early.

Data Checker sits at the top of each Einstein Prediction Builder page. It gives you a snapshot of your data and how many records you have at each stage of the build process. If you need to, adjust your settings for more records before building.

Check your data at these milestones when you build a prediction.

After you select an object to predict and segment your dataset.
After you build a filter-based prediction and define yes and no example values (example set).
After you define your records to predict (prediction set).
On the review page before you build.

How Data Checker Works
Status	What Data Checker Does	Example	Action
Success	After saving your settings, Data Checker shows a check mark. It checks your object, segment, example records, and records to predict, and finds that you have enough unique records on your object.	You’re predicting the likelihood that an opportunity closes by the end of the year. Your dataset has 1,000 Opportunity object records and 800 Yes and No Example records that Einstein can learn from. It has 400 records to predict. Because you met the minimum record count for each stage of the build process, Data Checker indicates that it can build successfully.	Your current data settings are sufficient to build the prediction.
Unknown	After saving your settings, Data Checker was unable to get an accurate record count. You can still build a prediction, or you can contact Salesforce Support to resolve the issue.	You add complex filter conditions, which narrow your Yes and No example records. After applying too many filters, you end up with a narrow dataset and don’t have enough object records.	Build your prediction. If Data Checker can’t evaluate whether you have sufficient data, an error appears in the scorecard after modeling is complete.
Error	After saving your settings, Data Checker found that your object didn’t have enough records to build a prediction.	You have enough total example records, but you don’t have enough Yes example records.	Adjust your example records so that you meet the minimum record requirement to build a prediction.