How to use parameter weight_column_name in AutoMLConfig Class for Automated Azure Machine Learning UI

Question

Hi everybody,

I'm aware using the SDK you can use weight_column_Name in the AutomLconfig, however I'm trying to add a weights column in the UI. How can I achieve this?

Kind Regards

Answer 1

I think the ui just has no path to move weight_column_name into the training graph… it never reaches the backend. run the job once from sdk with the weight set; if that still fails, check the column dtype and nulls, that’s all u need.

Answer 2

In Azure Machine Learning AutoML, the weight column is exposed as the weightColumnName (or WeightColumnName) property on the AutoML task configuration. This is supported across SDKs for table-based AutoML tasks (classification, regression, forecasting), but the context only documents programmatic usage, not a UI control in the studio.

Programmatic usage examples from the context:

• .NET (Resource Manager models):
  • For classification:

        var task = new ClassificationTask
        {
            WeightColumnName = "my_weight_column"
        };
    

    where WeightColumnName is defined as:

    The name of the sample weight column. Automated ML supports a weighted column as an input, causing rows in the data to be weighted up or down.

• Forecasting (Resource Manager models):

    var task = new MachineLearningForecasting
    {
        WeightColumnName = "my_weight_column"
    };
  

• PowerShell (Table vertical):

    public string WeightColumnName { get; set; }
  

  with the same description:

  The name of the sample weight column. Automated ML supports a weighted column as an input, causing rows in the data to be weighted up or down.

• JavaScript/TypeScript (TableVertical and derived tasks such as Classification, Regression, Forecasting):

    interface TableVertical {
      weightColumnName?: string;
    }
    
    const task: Classification = {
      weightColumnName: "my_weight_column",
      // ...other properties
    };
  

• Java (Classification):

    Classification classification = new Classification()
        .withWeightColumnName("my_weight_column");
  

  where withWeightColumnName is documented as:

  Set the weightColumnName property: The name of the sample weight column. Automated ML supports a weighted column as an input, causing rows in the data to be weighted up or down.

The provided context only describes how to set weightColumnName/WeightColumnName via SDKs and ARM/PowerShell models and does not document any Azure Machine Learning studio (UI) option to configure the weight column directly. Based on the context, configuring the weight column is done programmatically, not via a documented UI control.

---

References:

• ClassificationTask.WeightColumnName Property
• MachineLearningForecasting.WeightColumnName Property
• ITableVertical.WeightColumnName Property
• TableVertical interface
• Forecasting interface
• Classification Class
• Classification interface
• Regression interface