12 Blending gradient boosting and deep learning

This chapter covers

  • A review of the end-to-end gradient boosting example from chapter 7
  • A comparison of the results of the gradient boosting example from chapter 7 with a deep learning solution for the same problem
  • The result of ensembling a gradient boosted model with a deep learning model

In chapter 7, we did an in-depth exploration of an end-to-end example of using gradient boosting. We explored a dataset of Airbnb listings for Tokyo, we engineered features suitable for a pricing regression task, and then we created a baseline model trained on this dataset to predict prices. Finally, applying the techniques we had learned in the book up to that point, we optimized an XGBoost model trained on this dataset and examined some approaches to explain the behavior of the model.

12.1 Review of the gradient boosting solution from chapter 7

12.2 Selecting a deep learning solution

12.3 Selected deep learning solution to the Tokyo Airbnb problem

12.4 Comparing the XGBoost and fastai solutions to the Tokyo Airbnb problem

12.5 Ensembling the two solutions to the Tokyo Airbnb problem

12.6 Overall comparison of gradient boosting and deep learning

Summary