Standard CART decision trees are unstable—small changes in training data can produce substantially different tree structures and predictions. stable-cart addresses this by trading a small amount of accuracy for lower cross-bootstrap prediction variance through techniques like honest estimation (using separate data subsets for learning tree structure vs. estimating leaf values), lookahead search (considering multiple future splits before committing rather than making greedy single-step decisions), and bootstrap-aware split selection (penalizing or filtering out splits that are unstable across resampled datasets).
Cite: Gaurav Sood. "stable-cart: Lower Bootstrap Prediction Variance CART." https://gsood.com/research/stable-cart-lower-bootstrap-prediction-variance-cart/
BibTeX:
@software{soodstable,
author = {Gaurav Sood},
title = {{stable-cart: Lower Bootstrap Prediction Variance CART}},
url = {https://github.com/finite-sample/stable-cart}
}