Training Machine Learning Models in AMLS

Training Machine Learning (ML) models in Azure Machine Learning (AML) is key to enabling your data science workload. Typically, during the model creation process, data is split into test and training datasets. Models are then built with the training data and evaluated using the test dataset. During this process, many algorithms are selected and used to answer the question: what model will provide the best results on an unseen dataset? AML has the capability to log metrics, taking snapshots of the code that produced a given model performance to enable answering this question. AML comes with a variety of accelerating capacities. During this chapter, we will focus on the creation of experiments to train models and on the basic functionality of AML experiments to unlock using compute instances, compute clusters, and registered datasets.

Model training can be established through the AML Python SDK or the designer for a low-code experience. During...