Decision Trees — Applied Machine Learning in Python

In pre-pruning, you restrict while you’re growing. In post-pruning, you build the whole tree (greedily) and then you possibly merge notes. So you possibly undo sort of bad decisions that you made, but you don’t restructure the tree.

In scikit-learn right now, there’s only pre-pruning. The most commonly used criteria are the maximum depth of the tree, the maximum number of leaf nodes, the minimum samples in a node split and minimum decrease impurity.