Machine Learning Definitions and Concepts

This chapter offers a high-level definition and explanation of the machine learning concepts needed to use the Amazon Machine Learning (Amazon ML) service and fully understand how it works. The chapter has three specific goals:

• Listing the main techniques to improve the quality of predictions used when dealing with raw data. You will learn how to deal with the most common types of data problems. Some of these techniques are available in Amazon ML, while others aren't.
• Presenting the predictive analytics workflow and introducing the concept of cross validation or how to split your data to train and test your models.
• Showing how to detect poor performance of your model and presenting strategies to improve these performances.

The reader will learn the following:

• How to spot common problems and anomalies within a given dataset
• How to extract the most information out of a dataset in order to build robust models
• How to detect and improve upon poor predictive performance