Depth vs training / validation error

This demo sweeps tree depth on a fixed train/validation split and plots how classification error changes. The noisy moons and noisy ring datasets often show the clearest “train down, validation down then up” pattern.