Abstract
<jats:p>This article explores the integration of Brute Force (BF) methods and Greedy algorithms to develop a versatile decision tree (DT) algorithm that balances efficiency, complexity, and accuracy. The study introduces a parameterized search procedure that harnesses the strengths of both approaches, applicable to ensemble methods like AdaBoost and Random Forests (RFs). The primary research question investigates whether enhanced utilization of BF approaches can improve the quality of greedy DT algorithms without increasing their complexity. Key hypotheses address the trade-off between tree complexity and user interpretability, the potential for reduced complexity through BF methods, and the impact on computational time and accuracy. A qualitative field study supports these hypotheses, revealing that simpler DT structures are easier to interpret. Additionally, algorithmic evaluations suggest that incorporating BF methods can enhance the accuracy of greedy DTs without significantly extending computation time.</jats:p>