In this chapter, we explored the final step in the Amazon ML workflow, the predictions. Amazon ML offers several ways to apply your models to new datasets in order to make predictions. Batch mode involves submitting all the new data at once to the model and returning the actual predictions in a csv file on S3. Real-time predictions, on the other hand, are based on sending samples one by one to an API and getting prediction results in return. We looked at how to create an API on the Amazon ML platform. We also started using the command line and the Python SDK to interact with the Amazon ML service -- something we will explore in more depth in Chapter 7, Command Line and SDK.
As explained in the previous chapters, the Amazon ML service ...
