Azure ML Training Tool

Use the Azure Machine Learning Training tool to send data directly from an Alteryx workflow to the Azure Automated Machine Learning service. The Azure Machine Learning Training Tool starts the automated machine learning (ML) experimentation process to train and identify the best performing machine learning model.

To use this tool, sign up for a Microsoft Azure account.

For more information, see the Microsoft Azure Machine Learning documentation.

Configure the tool

Experiment setup

• Experiment name: Provide a name for the experiment you want to run. The experiment name must start with a letter and cannot contain spaces or special characters.
  Note that all individual runs with the same experiment name will be saved as the same experiment.
• Target variable: Select a target variable from your data stream - this is what you want to predict.
• Experiment type: Depending on your target variable, select classification or regression. Automated machine learning (automated ML), which is part of Azure Machine Learning, will try more than one algorithm for each experiment type.
  • Compute configuration: Select a compute configuration. This configuration impacts the resources used to run your experiment.
  • Compute name: The name of your remote compute target. A new compute will be created if it does not already exist.
  • CPU or GPU: Select the processor type you would like your experiment to be run on.
  • Min nodes: Specify the minimum number of nodes you would like to run your experiment on.
  • Max nodes: Specify the maximum number of nodes you would like to run your experiment on.
  • Max cores per iteration: The maximum number of cores to use in your experiment.
  • Max time per iteration: The maximum time, in minutes, that each iteration can run before being terminated.
  • Max experiment run time: The maximum time, in minutes, that the experiment can run before being terminated.

• Advanced options:
  • Number of iterations: The total number of algorithm and parameter combinations to test during your experiment.
  • Number of cross-validations: The number of cross-validations performed on the model.
  • Maximum concurrent iterations: The maximum iterations to be executed concurrently. This should be set to a value less than the number of cores specified in Compute configuration.