Azure-docs: No module named 'azureml.datacollector'

[ghost]

When creating an online (so not in local mode) service from an image with the data collector, I get the error:
from azureml.datacollector import ModelDataCollector
ImportError: No module named 'azureml.datacollector'

This is strange as in local mode I do not get this error (because I installed the package with pip).
Is data collection available?

---

Document details

⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

• ID: d9275aa2-0bee-c4a3-c710-fd3ffa2d200e
• Version Independent ID: 74cfa7b8-6859-1749-8b5c-1f8c29a404fe
• Content: Azure Machine Learning Model Data Collection API reference
• Content Source: articles/machine-learning/desktop-workbench/model-data-collection-api-reference.md
• Service: machine-learning
• Product: unspecified
• GitHub Login: @aashishb
• Microsoft Alias: aashishb

[YutongTie-MSFT]

@wernerstinckens Thanks for the feedback! We are currently investigating and will update you shortly.

[YutongTie-MSFT]

@wernerstinckens Thanks for the feedback! I have assigned the issue to the content author to investigate further and update the document as appropriate.

@aashishb Hi, can you please help to make sure if the data collector is available online? And if necessary, can you please update the document as well for users creating online service? Thanks.

[marthalc]

Hi @wernerstinckens we have discovered the root cause of the issue. Will get back to you as soon as the fix is ready. Thanks so much for your patience.
-Martha.

[angusmack-msft]

@marthalc we are also seeing the same error. We are on an AI Bootcamp training course and stuck on this part. Any updates please?

[marthalc]

@angusmack-msft and @wernerstinckens the fix should be available soon. We deploy by region, could you let me know in which region are you deploying on so we can prioritize it?

[angusmack-msft]

Hi @marthalc - I was using this in West Europe.

However, I manged to fix this - right at the end of the course on Friday - and I simply needed to include both the model management and data collector libraries. This was my working conda_dependencies.yaml file.

name: project_environment
dependencies:
 - python=3.5.2
 - scikit-learn=0.18.1
 - numpy=1.11.3
 - pip:
   # Required packages for AzureML execution, history, and data preparation.
   - --index-url https://azuremldownloads.azureedge.net/python-repository/preview
   - --extra-index-url https://pypi.python.org/simple
   - azureml-requirements

   # The API for Azure Machine Learning Model Management Service.
   # Details: https://github.com/Azure/Machine-Learning-Operationalization
   - azure-ml-api-sdk==0.1.0a11
   - azure-storage-blob

   #data collection
   - azureml-model-management-sdk==1.0.1b6.post1
   - azureml.datacollector==0.1.0a13

If there are other issues then please let us know and what the fix is.

Thanks.

[ghost]

West Europe.

[Gadnief]

Even after including the libraries in the yaml i get the ImportError when i try to deploy the model.pkl and scorer.py as a docker container to use it on edge.

[marthalc]

@Gadnief - Data Collection is currently not supported on edge devices. To get around the issue for deployment you simply need to remove the import and the calls to ModelDataCollector in your code.

[marthalc]

@wernerstinckens and @angusmack-msft the fix has been deployed. Please do let me know if you are still facing any issues. Thanks again for your patience and support.

[ghost]

@marthalc: it works fine now, tnx!

[marthalc]

@YutongTie-MSFT could we close this please?

[j-martens]

@YutongTie-MSFT #please-close

[ashoksamal6363]

is this available in north europe? i am getting this issue.

[haakar]

@ashoksamal6363 we moved model data collector from azureml.datacollector to azureml-monitoring on May 2018. Could you update your setup code and scoring script to use azureml-monitoring instead? To answer your question, yes, it is available in North Europe.
https://pypi.org/project/azureml-monitoring/

[mkrdip]

@haakar Have a look at this documentation.

The Azure Machine Learning Monitoring SDK will be retired soon .....

[ManojBableshwar]

Hi Mrinal - we have removed the message that talked about retiring model data collector. Model data collector is the recommended option to log model inference data in AzureML.

[mkrdip]

Does it makes sense to deploy a service in production with a package that is under development or testing phase? https://pypi.org/project/azureml-monitoring/

[mkrdip]

It'll be retiring soon says again here: https://docs.microsoft.com/en-us/python/api/overview/azure/monitoring/intro?view=azureml-monitoring-py