Appendix A: More Feature Engineering Operators
This appendix introduces a few more feature engineering operators that you can use in your chains when working with tabular and/or time series data and digs more into some of those already introduced in the course of the book. They’ve been omitted from the main chapters because their use is more sporadic, but they’re still valuable and can greatly improve the quality of your model’s predictions.
Feature Scaling
This section lists more methods available to perform feature scaling. As a recall, feature scaling is used to alter the domain of the input data, either with a linear transformation (min-max, z-score, robust operators), a non-linear one (Box-Cox and Yeo-Johnson), or an instance-based one (norm). ...