So far we have covered a lot of data science models, we talked about many supervised and unsupervised learning methods, including deep learning and XGBoost, and discussed how we can apply these models to text and graph data.
In terms of the CRISP-DM methodology, we mostly covered the modeling part so far. But there are other important parts we have not yet discussed: evaluation and deployment. These steps are quite important in the application lifecycle, because the models we create should be useful for the business and bring value, and the only way to achieve that is integrate them into the application (the deployment part) and make sure they indeed are useful (the evaluation part).
In this last chapter of the book we will cover exactly these missing parts--we will see how we can deploy data science models so they can be used by other services of the application. In addition to that, we will also see how to perform an online evaluation of already...