Summary

In this chapter, we have covered a lot about building predictive models using an Online Purchase dataset. We have explored two different tree-based models, random forest and GBDT, and how to build predictive models to forecast who is likely to convert. Using the same example, we have also discussed how we can build neural network models that are the backbone of deep learning models. There is great flexibility in how you architect the neural network model, such as wide network, deep network, or wide and deep network. We have briefly touched on the activation functions and optimizers while building neural network models, but we suggest you do some more in-depth research into how they affect the performances of neural network models. Lastly, we have discussed what A/B testing is, how to conduct A/B testing, and how to interpret the A/B testing results. We have simulated A/B testing with the models we built for a scenario where we want to choose the best model for capturing the...