Summary

In this introductory chapter, we presented the techniques used by the Amazon ML service. Although Amazon ML offers fewer features than other machine learning workflows, Amazon ML is built on a solid ground, with a simple yet very efficient algorithm driving its predictions.

Amazon ML does not offer to solve any type of automated learning problems and will not be adequate in some contexts and some datasets. However, its simple approach and design will be sufficient for many predictive analytics projects, on the condition that the initial dataset is properly preprocessed and contains relevant signals on which predictions can be made.

In Chapter 2, Machine Learning Definitions and Concepts, we will pe further into techniques and concepts used in predictive analytics.

More precisely, we will present the most common techniques used to improve the quality of raw data; we will spot and correct common anomalies within a dataset; we will learn how to train and validate a predictive model and how to improve the predictions when faced with poor predictive performance.