How to bypass ID column without being used in the training model but have it as output - Azure ML

Question

The input data in the model includes column ControlNo.

But I don't want this column being part of learning process so I'm using Select Columns in Dataset to exclude ControlNo column.

But as a output I want those columns:

ControlNo, Score Label, Score Probability

So basically I need NOT to include column ControlNo into learning process, but have it as output along with Score Label column.

How can I do that?

Answer 1

You have two options to do this, one less elegant than the other:

1. You can drop whatever column you don't want to use and add them together after getting the scores. Though you might need to reorder columns etc.

Or...

2. There is a block called Edit Metadata which can be used to do what you need. Select whatever columns from the column selector you don't want to use and use the option Fields > Clear Feature from the block properties. These columns won't be used in the calculations but will be present when you visualise the Score Model output together with the rest of the columns and the scores. Moreover, columns in that output statistics will be labelled as Feature type if they were used in the calculations. So you straightaway know which columns were actually used in the calculations or not. Block help is quite useful here.