Automl fails to prep data

Question

Automl fails to prep data

Kush Shah 0

I have a dataset that is mostly curated of ~30 features (25 decimal 5 numeric) and about 40k rows. In azure ml studio, In a Jupyter notebook from the UI I am able to run a random forest model on this data as expected but when I try to run this in automl with featurization enabled or disabled and with feature type set to auto orspecified the job fails or runs with "setting up the run" status till it is timed out. Below is the error from the child job:

Below is from the raw JSON:

error": {
        "error": {
            "code": "System",
            "severity": null,
            "message": "",
            "messageFormat": "Failed to get data from DataPrep. Exception Type: <class 'NotImplementedError'>",
            "messageParameters": {},
            "referenceCode": "prepare",
            "detailsUri": "https://aka.ms/azureml-run-troubleshooting",
            "target": null,
            "details": [],
            "innerError": null,
            "debugInfo": null,
            "additionalInfo": null
        }

Below is from the log:

File "/azureml-envs/azureml-automl/lib/python3.8/site-packages/azureml/dataprep/api/engineless_dataflow.py", line 90, in add_step
    raise NotImplementedError
azureml.train.automl.exceptions.ClientException: ClientException:
	Message: 
	InnerException: NotImplementedError: 
	ErrorResponse 
{
    "error": {
        "code": "System",
        "message": "",
        "reference_code": "prepare"
    }
}

I thought it was an issue with my data but I was able to get expected results running this locally and within the notebook in the studio. I tried to create a new compute instance as well but the same error persisted. Is there a step I am missing to setup a run on automl? I followed the steps in the UI as mentioned here: https://learn.microsoft.com/en-us/azure/machine-learning/tutorial-first-experiment-automated-ml?view=azureml-api-2

Ramr-msft 16,141 Reputation points

2023-10-30T13:32:33.3966667+00:00
@Kush Shah Thanks for the question, Can you please share the sample data to check from the AutoML UI. You can try the following that can help.

Check Your Data Types: Make sure the data types of your features are correctly identified in your dataset. Mismatched or incorrect data types can sometimes cause issues during the data preparation step.

Check Your Data Preparation Steps: If you’re using any custom data preparation steps, make sure they are correctly implemented and compatible with Azure Machine Learning.

Ramr-msft 16,141 Reputation points

2023-10-30T13:32:33.3966667+00:00

@Kush Shah Thanks for the question, Can you please share the sample data to check from the AutoML UI. You can try the following that can help.

Check Your Data Types: Make sure the data types of your features are correctly identified in your dataset. Mismatched or incorrect data types can sometimes cause issues during the data preparation step.

Check Your Data Preparation Steps: If you’re using any custom data preparation steps, make sure they are correctly implemented and compatible with Azure Machine Learning.

Automl fails to prep data

Activity