A real-time endpoint in Azure Machine Learning is specifically designed for real-time (or online) inference. It deploys a model as a web service that can receive data and return predictions synchronously with low latency. This deployment type is ideal for applications that require immediate responses, such as interactive websites, mobile apps, or fraud detection systems. The endpoint can be scaled to handle varying request loads.

A batch endpoint is used for asynchronous, long-running inference on large volumes of data, where immediate responses are not required.

A scheduled pipeline is used to automate recurring machine learning tasks, such as model retraining or batch scoring, not for serving real-time requests.

A data store is a reference to a storage service (like Azure Blob Storage) within an Azure ML workspace; it holds data, it does not serve models.

1. Microsoft. (2023). What are Azure Machine Learning endpoints? Microsoft Learn. Retrieved from https://learn.microsoft.com/en-us/azure/machine-learning/concept-endpoints. (See the "Online endpoints" section, Paragraph 1, which states they are for "real-time inference").

2. Microsoft. (2023). Deploy and score a machine learning model by using an online endpoint. Microsoft Learn. Retrieved from https://learn.microsoft.com/en-us/azure/machine-learning/how-to-deploy-online-endpoints. (See the "Introduction" section, Paragraph 1).