As artificial intelligence and big data have become a ubiquitous part of our everyday lives, cloud-based machine learning services are part of a rising billion-dollar industry. Among the several services currently available on the market, Amazon Machine Learning stands out for its simplicity. Amazon Machine Learning was launched in April 2015 with a clear goal of lowering the barrier to predictive analytics by offering a service accessible to companies without the need for highly skilled technical resources.

This introductory chapter is a general presentation of the Amazon Machine Learning service and the types of predictive analytics problems it can solve. The Amazon Machine Learning platform distinguishes itself by its simplicity and straightforwardness. However, simplicity often implies that hard choices have been made. We explain what was sacrificed, why these choices make sense, and how the resulting simplicity can be extended with other services in the rich data-focused AWS ecosystem.

We explore what types of predictive analytics projects the Amazon Machine Learning platform can address and how it uses a simple linear model for regression and classification problems. Before starting a predictive analytics project, it is important to understand what context is appropriate and what constitutes good results. We present the context for successful predictions with Amazon Machine Learning (Amazon ML).

The reader will understand what sort of problems Amazon ML can address and the assumptions with regard to the underlying data. We show how Amazon ML solves linear regression and classification problems with a simple linear model and why that makes sense. Finally, we present the limitations of the platform.

This chapter addresses the following topics: