Data classification: Predicted failed with error "Column 'Probability' not found" for Microsoft SQL Server data source when using metric "F1".

[v-Hailishi]

System Information (please complete the following information):
Windows OS: Windows-11-Enterprise-22H2
ML.Net Model Builder 2022: 17.14.4.2317701 (Main Build)
Microsoft Visual Studio Enterprise: 2022(17.5.2)
.Net: 6.0,7.0

Describe the bug

• On which step of the process did you run into an issue:
  Predict failed with error "Column 'Probability' not found" for Microsoft SQL Server data source when using metric "F1".

TestMatrix
https://testpass.blob.core.windows.net/test-pass-data/wikipedia-detox-250-line-data.tsv

To Reproduce
Steps to reproduce the behavior:

1. Open Visual Studio app.
2. Select Create a new project from the Visual Studio start window.
3. Choose the C# Console App (.NET Core) project template.
4. Add model builder by right click on the project.
5. Select "Data classification scenario.
6. On Data page, choose the "Microsoft SQL Server" data source and .
7. Click the "Advanced data options..." link, go to the "Validation data" tab, choose the second option "Split (Training/Validation)".
8. On the Train page, choose "F1" metric and keep "60" seconds training time.
9. Click "Start training" button to completed training.
10. If the best model trainer is "FastForestBinary", then go to Evaluate page, the prediction would be failed.

Expected behavior
Prediction should be successful.

Screenshot

Additional context
Only if the following 4 conditions are met at the same time, I can repro this bug:

1. Select the "Split (Training/Validation)" validation data strategy.
2. Select "F1" metric.
3. Keep "60" seconds training time.
4. The best model trainer should be "FastForestBinary".
   

[LittleLittleCloud]

Probability only exists for calibrated trainer. But not all trainers in AutoML has been calibrated. I'll push a fix

[v-Hailishi]

The bug has been fixed on the latest main build 17.15.0.2320602.