NCART: Neural Classification and Regression Tree for tabular data

Jiaqi Luo, Shixin Xu

Abstract: Highlights•We propose a neural network called Neural Classification and Regression Tree (NCART) for tabular data.•NCART exhibits interpretability, generating feature importance similar to those produced by decision trees.•The model is applicable to datasets of varying scales while maintaining lower computational costs.•NCART demonstrates superior performance compared to existing deep learning models and remains competitive with decision tree models.