Decision Tree Pruning Method Using Delayed Sampling
S. A. Mitrofanov, Eugene Stanislavovich Semenkin · 2024
Algorithms based on decision trees have been around for quite some time, however, they are still one of the most popular because of their efficiency and the interpretability of their results. One of the main problems of decision trees is their high tendency to overfit. There are various ways to solve this problem, but this paper proposes a new method based on tree pruning using delayed sampling. The tree is trained using the Separation Measure and Differential Evolution algorithms. The proposed approach improves the standard method and is compared with it. The proposed approach demonstrates higher efficiency than the standard algorithm, and also improves the interpretability of the resulting trees.