I use random forests as a first model for basically everything classification or regression.

This looks incredibly useful for understanding samples of individual trees, for which I use the nice but limited inbuilt tree object in sklearn.

Thank you!