Defining spatial feature engineering

As we mentioned at the beginning of this chapter, feature engineering refers to the manipulation and transformation of raw data into features that are best suited to your analytical exercise. In data science, feature engineering can take many forms, including the following:

• Filling missing values, leveraging expert intuition, or various machine learning-based approaches
• Scaling and normalization, whereby the range and center of data are adjusted to help train models and allow easier interpretation later on
• Feature encoding, whereby categorical data is converted to binary True or False representation across multiple columns

Spatial feature engineering is very similar to the approaches taken in more general data science. It is the process of creating, or engineering, new and additional information from raw data using geographic context and knowledge. Engineering new features can be done by connecting data from two or more datasets...