Unable to test Machine Learning Endpoint, no sample JSON provided

Hi Damian McCracken

When deploying an Auto ML model as a real-time endpoint in Azure Machine Learning, it’s common to encounter issues where the test tab does not display a sample JSON input. This typically happens when the model metadata does not include a clearly defined input schema, leaving users uncertain about the expected format. The endpoint may deploy successfully, but without a sample payload, it’s unclear how to structure the input for testing. The expected format usually mirrors the structure of a single row from the training dataset. For example, if your dataset includes features like "age", "income", and "education", the input should be structured as:

{
  "input_data": {
    "data": [
      {
        "age": 35,
        "income": 55000,
        "education": "Bachelor"
      }
    ]
  }
}

To discover the correct schema, you should review the original model card in Azure ML Studio, which often includes details about the training data and feature engineering steps. You can also inspect the training dataset directly to identify column names and data types. If you're using the Azure ML SDK, you can retrieve model input details programmatically. Additionally, enabling logging or verbose output during test requests can help identify schema mismatches. If none of these methods work, consider redeploying the model with explicit input schema definitions or using a custom scoring script that validates incoming requests.

Reference links: Deploy and score a machine learning model by using an online endpoint , Deploy a model as an online endpoint

Hope it helps!

Thank you