Share via

Unable to Use Uploaded CSV (MLTable) Dataset in Azure AutoML

Scott Campbell 20 Reputation points
2025-05-06T19:03:53.4266667+00:00

Hi,

I'm attempting to use Azure Machine Learning's AutoML feature to train a classification model using a CSV file I've uploaded. I've followed all best practices, including:

  • Uploading the CSV via the Data > +Create > Data Asset path
  • Selecting Table (mltable) as the dataset type - including a valid, properly formatted MLTable YAML file in the same folder
  • Uploading the folder (containing both xxx.csv and MLTable)
  • Successfully passing Azure's upload and validation steps (including "Skip validation" where necessary)

Despite all of this, when I attempt to use the dataset within the Automated ML job wizard, I consistently encounter a "Invalid MLTable specified" error when Azure attempts to preview or validate the dataset. (When trying just the CSV, it doesn't list my File option as a valid data type dropdown...only mltable or tabular). For the MLTable option, I'll get through to the end of the process and it will say, "Error loading data preview".

This error persists even when:

Skipping data validation

Uploading various cleaned versions of the file

Following Microsoft’s documentation and community guidance precisely

Something seems seriously wrong for such a relatively simple data loading process. I need help understanding why I can't use a CSV file and why a valid MLTable dataset is being rejected by the AutoML pipeline. Can someone please help? How I can resolve this to proceed?

Thank you!

Scott Campbell

Azure Machine Learning
Azure Machine Learning
An Azure machine learning service for building and deploying models.
3,259 questions
0 comments
Accepted answer
  1. Pavankumar Purilla 7,015 Reputation points Microsoft External Staff Moderator
    2025-05-06T22:01:46.28+00:00

    Hi Scott Campbell,
    I'm glad that you were able to resolve your issue and thank you for posting your solution so that others experiencing the same thing can easily reference this!

    Since the Microsoft Q&A community has a policy that "The question author cannot accept their own answer. They can only accept answers by others", I'll repost your solution.

    Please click "Accept" the answer as original posters help the community find answers faster by identifying the correct answer.

    Issue : Unable to Use Uploaded CSV (MLTable) Dataset in Azure AutoML

    Resolution : The key was to stop trying to preview the data and instead use the "Tabular" dataset type throughout the entire process.

    Here’s what worked:

    I uploaded the CSV using the +Create > Data Asset path.

    I selected Tabular as the data type, not File (uri_file) or Table (mltable).

    Once the data asset was created, it appeared properly in the AutoML wizard.

    I selected my target column, created a compute cluster, and ran the experiment successfully.

    Turns out, Tabular datasets are fully supported for CSV files in AutoML, and most of the older documentation referring to MLTable formatting wasn’t necessary in my case.

    Please don’t forget to close the thread by clicking "Accept the answer" wherever the information provided helps you, as this can be beneficial to other community members.

    Accept answers on Microsoft Q&A | Microsoft Learn

    An accepted answer is the answer that the person who asked the question chooses as the one they think best solves their problem.

1 additional answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. Scott Campbell 20 Reputation points
    2025-05-06T21:39:33.2333333+00:00

    Thanks to some experimentation and updated guidance, I was able to resolve this.

    The key was to stop trying to preview the data and instead use the "Tabular" dataset type throughout the entire process.

    Here’s what worked:

    I uploaded the CSV using the +Create > Data Asset path.

    I selected Tabular as the data type, not File (uri_file) or Table (mltable).

    Once the data asset was created, it appeared properly in the AutoML wizard.

    I selected my target column, created a compute cluster, and ran the experiment successfully.

    Turns out, Tabular datasets are fully supported for CSV files in AutoML, and most of the older documentation referring to MLTable formatting wasn’t necessary in my case.

    Thanks for the help, and I hope this helps someone else avoid the same frustration!

    — Scott Campbell

    0 comments

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.