Chapter 10: Feature Transformation and Scaling
In the previous chapter, you have seen how to manage drift and drift detection in streaming and online machine learning models. Drift detection, although not the main concept in machine learning, is a very important accessory aspect of machine learning in production.
Although many secondary topics are important in machine learning, some of the accessory topics are especially important with online models. Drift detection is particularly important, as the model's autonomy in relearning makes it slightly more black-box to the developer or data scientist. This has great advantages only as long as the retraining process is correctly managed by drift detection and comparable methods.
In this chapter, you will see another secondary machine learning topic that has important implications for online machine learning and streaming. Feature transformation and scaling are practices that are relatively well defined in traditional, batch machine...
