Which Amazon SageMaker method is used for making predictions on large datasets in bulk?

The Batch Transform method in Amazon SageMaker is specifically designed for making predictions on large datasets in bulk. This functionality allows users to input a dataset and receive predictions for all the data at once, which is especially useful when working with a vast amount of information that cannot be processed in real-time.

Batch Transform is useful when the latency of predictions is not a critical factor, as it can operate on multiple data points simultaneously. Users typically provide a location for their input data stored in Amazon S3, specify the output path for the predictions, and define the appropriate model to use. SageMaker handles the deployment of the model and optimizes the resources needed to process the large dataset efficiently.

In contrast, real-time inference focuses on making predictions for individual data points as they arrive, which is more suited for use cases requiring immediate responses. Stream processing generally involves continuously processing data streams in real-time, and instant prediction is not a recognized term in Amazon SageMaker related to bulk predictions.