EINSTEIN_MODEL.md

Set up model in Einstein Studio

1. Log in to the org

2. Navigate to ML Workspace / Einstein Studio

3. Select ML Workspace

4. Click New

You should see a toast message that the end point was saved successfully

5. Give your model a name and click create

6. Select Endpoint

7. Click on add endpoint

8. Enter inference url from Databricks as well as request format as dataframe split

9. Select Authentication type, Auth Header= "Authorization"

10. Secret Key = "Bearer <>"

11. Click Save.

You should see a toast message that the end point was saved successfully

12. Select input features

13. Click on Add input features

14. Choose the DMO

Choose the DMO that has all the fields for model scoring in this case it is account contact DMO.

15. Click Save

16. Select fields from DMO for scoring

Now start selecting the fields from the DMO for model scoring. Note that the feature API name of the field selected should match the names the model is expecting for instance as shown in the query endpoint dialog above

17. Drag each predictor and click done one by one in the specific order

18. Once you enter all the predictors in the click on save

19. Next go to output Predictions

20. Give the DMO a name.

This is where the output predictions will be saved

21. Click save

22. Enter the outcome variable API name and the json key

Note that in this case the json key is - $.predictions

23. Click Save

24. Now activate the model

25. Once model is activated refresh it to see the predictions in the DMO