Machine learning models learn patterns from data. However, the raw data collected often isn't in the ideal format for these models to learn effectively. The process of selecting, transforming, and creating the input variables, known as features, from raw data is called feature engineering. This initial chapter establishes the foundation for understanding this process.

You will begin by situating feature engineering within the broader machine learning workflow. We will define what constitutes a 'feature' and examine how the quality and relevance of features directly influence a model's ability to learn and generalize. We will also look at the common types of data encountered, such as numerical and categorical data, and the specific considerations they require. Finally, this chapter provides a high-level overview of the main tasks involved in feature engineering, setting the stage for the techniques detailed in subsequent chapters.