The Matheuristic for Building Cost-Constrained Decision Trees with Multiple Condition Attributes

Cost factors are frequently essential in many real-world applications. Many previous studies in machine learning included costs, particularly when creating decision tree models. This research also takes into account a cost-sensitive decision tree construction problem, with the premise that test costs must be spent in order to get the values of the decision attribute and that a record must be categorized without surpassing the expenditure cost threshold. Furthermore, our problem handles records with multiple condition attributes. A mathematical programming heuristic based on Variable Neighborhood Descent (VND) is introduced to compare to the existing approach in the literature. The experimental results show that our approach not only satisfactorily handles small and medium datasets with multiple condition attributes under different cost constraints but also outperforms the existing method.