LDT: Layered decision tree based on data clustering

Decision tree can be considered as one of the most widely used methods due to the acceptable accuracy and interpretable results. The main limitation of this method is uncontrolled growing of tree that leads to produce complex model and degrade comprehensibility. In this paper, we propose layered decision tree (LDT) approach based on data clustering. The proposed algorithm, initially cluster data and sort them with respect to their importance. Then, these data divided into groups that each of them used for construct one layer of tree. Finally, tree will be updated based on maximum depth of each layer. Practical results show that LDT outperform than popular methods, in the term of accuracy, processing time, and complexity.