CODEWITHSUNDEEP
azureazure-blob-storageazure-machine-learning-service
Martin Vaughan
Martin Vaughan

Reputation: 423

Azure Machine Learning - error trying to view uploaded data in tutorial

I have been trying to follow the Azure Machine Learning tutorial at https://microsoftlearning.github.io/mslearn-ai-fundamentals/Instructions/Labs/01-machine-learning.html. I have followed the instructions to the letter but encounter an error when I try to upload the data (a csv file and MTable file previously downloaded and extracted as per the tutorial instructions).

For the exact steps leading up to this, please see the instructions in the link above. When I get to the Upload folder step, I first get the error:

The provided path is not valid or the files could not be accessed. If you want to proceed, consider checking "Skip data validation" and click "Next." We will not validate your data path or query, or try to access your data for preview and schema checks.

Error: Request failed with status code 400
    at uO (https://ml.azure.com/assets/index-0dc6c834.js:210:4072)
    at lO (https://ml.azure.com/assets/index-0dc6c834.js:210:4257)
    at XMLHttpRequest.D (https://ml.azure.com/assets/index-0dc6c834.js:211:1665)

The JSON for the error is

{
  "error": {
    "code": "ScriptExecution.StreamAccess.NotFound",
    "message": "NotFound",
    "additionalInfo": null,
    "details": []
  },
  "correlation": {
    "operation": "8e4f6116536b48eb819e5e4dc286b10d",
    "request": "8cf3ab591664058e"
  }
}

The only way to proceed at this point is click the Skip data validation button. At this point, the tutorial instructs one to:

Select Create. After the dataset is created, select the bike-rentals dataset to continue to submit the Automated ML job.

However this results in new error:

Error loading data preview Not Found.

Error: Request failed with status code 400
    at uO (https://ml.azure.com/assets/index-0dc6c834.js:210:4072)
    at lO (https://ml.azure.com/assets/index-0dc6c834.js:210:4257)
    at XMLHttpRequest.D (https://ml.azure.com/assets/index-0dc6c834.js:211:1665)

The JSON is:

{
  "error": {
    "code": "ScriptExecution.StreamAccess.NotFound",
    "message": "NotFound",
    "additionalInfo": null,
    "details": []
  },
  "correlation": {
    "operation": "8e4f6116536b48eb819e5e4dc286b10d",
    "request": "8b755cebf4a8eeb6"
  }
}

Another user reports a similar error (different tutorial) here: Azure AutoML and Blob Access Issue. A Microsoft vendor advises:

To resolve the issue, please ensure that the path provided for the MLTable file is correct and that you have the necessary permissions and access rights to the Blob storage account and then follow the steps to create the Blob storage account correctly and try again.

However, neither I nor the other user in question knew how to do this - these are, after all, beginner tutorials!

Edit The content of the MLTable file is

# MLTable definition file

paths:
  - file: ./daily-bike-share.csv
transformations:
  - read_delimited:
        delimiter: ','
        encoding: 'ascii'

Can anyone please give clear advice as to how to resolve this issue?

Upvotes: 0

Views: 459

Answers (1)

JayashankarGS
JayashankarGS

Reputation: 8020

Actually, the file path provided in your MLTable file sometime gives you this kind of error.

Try altering your MLTable file like any one of the below.

paths:
  - file: "./daily-bike-share.csv"
transformations:
  - read_delimited:
        delimiter: ','
        encoding: 'ascii'

Or give the pattern

paths:
  - pattern: "*.csv"
transformations:
  - read_delimited:
        delimiter: ','
        encoding: 'ascii'

Or direct file name.

paths:
  - file: daily-bike-share.csv
transformations:
  - read_delimited:
        delimiter: ','
        encoding: 'ascii'

Actually, it should work for the current definition sometimes, but you get error because of the file path only.

Try above ways to resolve the issue.

Upvotes: 0

Related Questions