0
votes

I have written a Jupyter notebook in Azure Machine Learning which executes without issue. Now, I have to publish the code as a web service. "From Jupyter Notebook to Azure Web App in 5 Easy Steps" describes converting a notebook into webservice using Docker and Azure Container Registry

Is there any alternative simple method within azure machine learning to convert the notebook into a web service? This can be done in the designer section:

see screenshot.

1
what does the notebook do? make a model? score new data against a model? please include more information.Anders Swanson

1 Answers

1
votes

The "Designer" section of Azure ML is just a UI for creating Azure Machine Learning Pipelines. You can publish Pipelines as a "web service". Note that this is different than deploying a model as a webservice. Unfortunately, the Designer UI does not currently support NotebookRunnerStep, but the SDK does. So your options are:

  • Use the SDK to define an Azure ML Pipeline with a NotebookRunnerStep, or
  • turn your notebook into a Python script and use the Designer to create a Pipeline with a PythonScriptStep step