Exercises

1. In the decision tree click-through prediction project, can you also tweak other hyperparameters, such as min_samples_split and class_weight? What is the highest AUC you are able to achieve?
2. In the random forest-based click-through prediction project, can you also tweak other hyperparameters, such as min_samples_split, max_features, and n_estimators, in scikit-learn? What is the highest AUC you are able to achieve?
3. In the GBT-based click-through prediction project, what hyperparameters can you tweak? What is the highest AUC you are able to achieve? You can read https://xgboost.readthedocs.io/en/latest/python/python_api.html#module-xgboost.sklearn to figure it out.