Refactor Train to use Azure ML Pipelines

[stevehadd]

As part of scaling up our machine learning training, we are going to use AzureML compute clusters. The current preferred way to send jobs to an AzureML compute cluster is to create an Azure ML pipeline. This allows the code to be broken up into modular units, such as scaling, train/test split, training, evaluation, plotting etc. which are portable and reusable.

• https://docs.microsoft.com/en-us/azure/machine-learning/concept-ml-pipelines
• https://docs.microsoft.com/en-us/azure/machine-learning/how-to-create-machine-learning-pipelines
• https://www.coursera.org/projects/azure-machine-learning-studio-pipeline
• https://github.com/Azure/MachineLearningNotebooks/blob/master/how-to-use-azureml/machine-learning-pipelines/intro-to-pipelines/aml-pipelines-getting-started.ipynb

[stevehadd]

Definitely something to explore, maybe in conjunction with other projects