Summary

In this chapter, we reviewed the entire data science model-building process. We started with raw data and a somewhat vaguely defined use case. Further inspection of the data allowed us to refine the problem statement to one that was relevant to the business and that could be addressed with the data at hand. We performed extensive feature engineering in the hopes that some features might be important predictors in our model. We introduced an efficient and powerful method of model building using H2O AutoML to build an array of different models using multiple algorithms. Selecting one of those models, we demonstrated how to further refine the model with additional hyperparameter tuning using grid search. Throughout the model-building process, we used the diagnostics and model explanations introduced in Chapter 7, Understanding ML Models, to evaluate our ML model. After arriving at a suitable model, we showed the simple steps required to prepare for the enterprise deployment of...