In this chapter, we will learn about the concept of model performance deterioration using an example of ad-click conversion. Our goal is to identify ad-clicks that result in mobile app downloads. In this case, the ads are for marketing mobile apps.
To address the deterioration of model performance, we will learn about feedback loops, pipelines in which we retrain models as new data becomes available and assess model performance. Consequently, trained models are constantly kept up to date with the changing patterns in input or training data. The feedback loop is very important when it comes to making sound business decisions based on model output. If a trained model does not adequately capture patterns in dynamic data, it is likely to produce sub-optimal results.
We will cover the following topics in this chapter:
- Monitoring models...
