Chapter 7. Feature Engineering Part II

Introduction

We have recognized the importance of feature engineering. In the previous chapter, we discussed techniques that enable us to select from a range of features and work effectively to transform our original data into features, which can be effectively processed by the advanced ML algorithms that we have discussed thus far.

The adage garbage in, garbage out is relevant in this context. In earlier chapters, we have seen how image recognition and NLP tasks require carefully-prepared data. In this chapter, we will be looking at a more ubiquitous type of data: quantitative or categorical data that is collected from real-world applications.

Data of the type that we will be working with in this chapter is common to many contexts. We could be discussing telemetry data captured from sensors in a forest, game consoles, or financial transactions. We could be working with geological survey information or bioassay data collected through research...