Skip to content

Azure Machine Learning

Element Name: Azure Machine Learning

Category: Cognitive.

Description: Advanced extension module for the use of Azure Machine Learning (Azure ML), a cloud-based service to facilitate the development, training, and deployment of machine learning models.

Functionality: Integrate with the Azure Machine Learning (Azure ML) machine learning model created and trained in Azure to leverage the inference capabilities of that model for use in Lynn.

Prerequisites:

  • Create an Azure Machine Learning machine learning model. For more information, visit Creating a Model

  • Train the model: For more information, follow the Microsoft tutorial Train a Model

  • Publish the model: For more information, visit Deploy a model

Action Configuration Fields

[Name]: A unique descriptor for the configuration box used for identification within the workflow.

[Expression to evaluate]: A string-type field where you should insert the expression or entity containing the expression you want to evaluate..

[REST endpoint Basic consumption info]: A field where you should specify the endpoint for integration, which is obtained from the ml.azure.com (Machine Learning Studio) portal. Locate the created model and then its endpoint. Click on the Consume tab to find the endpoint URL for the connection:

[Authentication Primary key or Secondary key]:A field where you should specify the authentication key for integration, which is obtained from the ml.azure.com (Machine Learning Studio) portal. Locate the created model and then its endpoint. Click on the Consume tab to find two keys; it is preferable to use the first one:

[Entity where the evaluation result will be deposited]: A selection box where you can indicate or create the entity where integration data will be deposited.

[Intention to execute in case of an error. The action flow will be interrupted, and the error description will be passed to the selected intention with the name: 'MLazureError']: A dropdown list where you should select the entity name for error handling.

Basic Example of Azure Machine Learning Implementation

--Hypothetical Context: Integrating with a model that helps personalize recommendations and marketing strategies for customers visiting a website. In this context and according to the model's training, the service expects input data and returns customer purchase predictions.

--Prerequisites:

  • Azure Machine Learning Machine Learning Model Name: Web_Compras

  • Train the Model: The model should be trained with data such as demographic information, browsing behavior, and customer purchase history.

  • Publish the Model: The model will be published as a web service. The endpoint name would be: endpoint-online-WS-compras

--Input:

[Expression to evaluate]: ID_CLIENTE

[REST endpoint Basic consumption info]: https://endpoint-online-WS-compras.inference.ml.azure.con/score

[Authentication Primary key or Secondary key]: fTY789oTfRd346oO

[Entity where the evaluation result will be deposited]: RESULT_WS_COMPRAS

[Intention to execute in case of an error. The action flow will be interrupted, and the error description will be passed to the selected intention with the name: 'MLazureError']: ERROR_WS

--Resultado: RESULT_WS_COMPRAS <--

    {
        "option_1": Product_A,
        "option_2": Product_N,
        "option_3": Product_M,
        "option_4": Product_C
    }