Azure ML - Deployed to Inference Cluster throws 500 Server Error - MissingFeaturesError

Question

We have an Azure ML model we are ready to deploy to an http endpoint for consumption and testing.

We are using this tutorial (https://learn.microsoft.com/en-us/azure/machine-learning/tutorial-designer-automobile-price-deploy) as the jumping off point for deploying our ml model. We have created an inference cluster, converted the training model to a real-time inference model and deployed. Deployment looks successful. However, when testing (both via the Test tab in the Azure ML Workspace and via http POST) the server throws a 500. The MissingFeaturesError follows:

File "/azureml-envs/azureml_9b50686470a92ca74f0d62e2629faaec/lib/python3.6/site-packages/azureml/studio/modules/ml/common/base_learner.py", line 289, in _validate_no_missing_feature
ErrorMapping.throw(MissingFeaturesError(required_feature_name=';'.join(missing_feature_list)))
> missing_feature_list = ['Miles', 'Age', 'Gender', 'MarriagetPlans']

File "/azureml-envs/azureml_9b50686470a92ca74f0d62e2629faaec/lib/python3.6/site-packages/azureml/studio/common/error.py", line 814, in throw
raise err
> err = MissingFeaturesError('Features for Miles;Age;Gender;MarriagetPlans required but not provided.',)

MissingFeaturesError: Features for Miles;Age;Gender;MarriagetPlans required but not provided.

In both test cases (via Test tab in Azure and http POST to the endpoint) all the required data is indeed provided. The request body definitely includes Miles, Age, Gender, MarriagetPlans.

What is going on here?

Answer 1

You can add a additional tag "azure-machine-learning" to make more people know your problem, I believe it's problem of AML studio

Answer 2

bump Still a problem.