locked
Retraining model with Azure data factory RRS feed

  • Question

  • Hi,

     

    I'm working on retraining model.

    I'd read "Retrain a New Resource Manager based web service using the Machine Learning Management PowerShell cmdlets" 
    https://docs.microsoft.com/en-us/azure/machine-learning/machine-learning-retrain-new-web-service-using-powershell#update-the-web-service-with-new-web-service-definition
    and "Create predictive pipelines using Azure Machine Learning and Azure Data Factory"
    https://docs.microsoft.com/en-us/azure/data-factory/data-factory-azure-ml-batch-execution-activity#updating-models-using-update-resource-activity

    I have a question about the "update the scoring (prediction) web service".
    Is it the same thing that the "Update the web service with new Web Service Definition" of "Retrain a New Resource Manager based web service using the Machine Learning Management PowerShell cmdlets" and the "Updating models using Update Resource Activity" of "Create predictive pipelines using Azure Machine Learning and Azure Data Factory" ?

    Can I use the method of the method of PowerShell instead of the pipeline of "Update Resource Activity" ? If it is the same thing.
    I want to do is that, using PowerShell to update the web service with new Web Service Definition.
    In the definition json file. 
    The the uri of locationInfo of retrain model had been modified to new position of retrain model.
    So, everytime I retrain model. It'll always update the ".iLearner" file in the blob.
    Therefore, in data factory pipeline just only have "AzureMLBatchExecution" activity.

    So when pipeline schedule performed. 
    The model had been retrained.
    The prediction web service is call the retrained model.
    So prediction web service can directly use without "Update Resource Activity" in data factory pipeline.

    English is not my native language. So, I hopefully someone understand what I say. Thanks!


    Friday, June 2, 2017 1:52 AM