# Overview
This final block of the book shows how the ideas, tools and building blocks explored in the previous two blocks can be put together in a coherent way. We show this using a few specific applications. Part of our intention here is to take traditional themes in data science (un/supervised learning, feature engineering), and to "sprinkle a bit of geo dust" to show how the way of thinking we have advanced in so far in this book can be used to obtain further insight beyond what traditional methods offer. The other part of our intention here is to show examples of *genuinely geographic* data science. Together, this will make the rest of the knowledge in the book useful for most readers.
[Chapter 9](../../notebooks/09_spatial_inequality) presents an explicitly spatial perspectice on inequality. Be it economic, social or of other nature, inequality often manifests itself in an explicitly-geographical way. For isntance, inequality between people is often most clearly recognized as inequality *between places*. This chapter shows how one can apply a "geographic mind" to traditional, aspatial measures; in doing so, it also illustrates a broader approach to explicitly spatialise non-spatial measures that can be deployed in a variety of contexts. [Chapter 10](../../notebooks/10_clustering_and_regionalization) considers unsupervised statistical learning and puts it through the lens of Geography. We divide this in two sections: the non-spatial clustering of geographic units, and explicitly spatial algorithms of clustering, or regionalisation methods. [Chapter 11](../../notebooks/11_regression) moves from unsupervised to supervised learning. We consider the case of regression and present both the intuition and technicalities of building space into a regression framework as a "first-class" citizen. Finally, in [Chapter 12](../../notebooks/12_feature_engineering), we flip the approach and, instead of showing how to build space into the model, we explore how geography can be embedded into the data we feed our supervised algorithms. In the machine learning jargon, polishing data into a shape that can be used in modelling is called "feature engineering". Since we present techniques to do this taking advantage of Geography and spatial relationships, we call it "Spatial Feature Engineering".