Session Ready
Exercise

Random forests

Random Forests are a classic and powerful ensemble method that utilize individual decision trees via bootstrap aggregation (or bagging for short). Two main hyperparameters involved in this type of model are the number of trees, and the max depth of each tree. In this exercise, you will implement and evaluate a simple random forest classifier with some fixed hyperparameter values.

X_train, y_train, X_test, y_test are available in your workspace. pandas as pd, numpy as np, and sklearn are also available in your workspace. RandomForestClassifier() from sklearn.ensemble is available as well, along with roc_curve() and auc() from sklearn.metrics.

Instructions
100 XP
  • Create a random forest classifier with 50 trees, and a max depth of 5.
  • Train the classifier and get probability scores via .predict_proba(), and predictions via .predict() for the testing data.
  • Evaluate the AUC of the ROC curve for the classifier using first roc_curve() to calculate fpr and tpr, and then auc() on the result.
  • Evaluate the precision and recall for the classifier.