The output of running WebAPI is confusing.

[vzhuqin]

System Information (please complete the following information):

• Model Builder Version (available in Manage Extensions dialog): 16.9.1.2152802 (Main)
• Microsoft Visual Studio Enterprise 2019: 16.11.5

Describe the bug

• On which step of the process did you run into an issue: running the generated WebAPI projects
• Clear description of the problem: why print the Feature columns with weird numbers, are they all scores?

To Reproduce Steps to reproduce the behavior:

1. Select Create a new project from the Visual Studio 2019 start window;
2. Choose the C# Console App (.NET Core) project template with .Net 5.0;
3. Add model builder by right click on the project;
4. Select Data classification to complete training;
5. Add WebAPI project to solution on Consume page;
6. Run the WebAPI project and use powershell to run it;
7. See the output is very confusing.

Expected behavior A clear and concise description of what you expected to happen.

Screenshots If applicable, add screenshots to help explain your problem.

Additional context Add any other context about the problem here.

[LittleLittleCloud]

This is because we uses schema to generate model output class now, so it includes all columns rather than just score. And the confusing value is encoded data/vector from model output.

I personally prefer this kind of output because it accurately shows what you get from model. @JakeRadMSFT @briacht thoughts?

[beccamc]

Jake will take a look at this strange output. What is in the project we generate? Doesn't seem right.

[JakeRadMSFT]

This is because we uses schema to generate model output class now, so it includes all columns rather than just score. And the confusing value is encoded data/vector from model output.

I personally prefer this kind of output because it accurately shows what you get from model. @JakeRadMSFT @briacht thoughts?

I oddly don't remember that change ... I would have expected just the output.

@michaelgsharp has a way to get just the output columns ... I think ... from schema.

[michaelgsharp]

This will figure out the Score, Probability, and PredictedLabel columns.

var outputSchema = model.GetOutputSchema(data.Schema);
var columnType = default(ReadOnlyMemory<char>);
DataViewSchema.Column scoreColumn = default;
DataViewSchema.Column predictedLabelColumn = default;
DataViewSchema.Column probabilityColumn = default;

foreach (var column in oSchema.Reverse())
{
                try
                {
                                // This will throw an error if the annotation isn't found
                                column.Annotations.GetValue("ScoreValueKind", ref columnType);
                                // We only hit this code if the error isn't thrown.
                                if (columnType.ToString() == "Score")
                                                scoreColumn = column;
                                if (columnType.ToString() == "PredictedLabel")
                                                predictedLabelColumn = column;
                                if (columnType.ToString() == "Probability")
                                                probabilityColumn = column;
                }
                catch (InvalidOperationException)
                {
                                // Don't do anything here. This is just so it will continue 
                }
}

[michaelgsharp]

If it doesn't, it means that some of our trainers aren't annotating the columns correctly, and thats a bug we would need to fix.