Feature improvement is about recognizing areas of issue and improvement in our data and figuring out which cleaning methods will be the most effective. Our main takeaway should be to look at data with the eyes of a data scientist. Instead of immediately dropping rows/columns with problems, we should think about the best ways of fixing these problems. More often than not, our machine learning performance will thank us in the end.
This chapter contains several ways of dealing with issues with our quantitative columns. The next chapter will deal with the imputing of categorical columns, as well as how to introduce brand new features into the mix from existing features. We will be working with scikit-learn pipelines with a mix of numerical and categorical columns to really expand the types of data we can work with.
