MULTICRITERIA OPTIMIZATION AND ANALYSIS ON DECISION TREE-BASED MACHINE LEARNING MODELS

Machine learning models have emerged as essential tools in the landscape of large corporations. These models act as robust analytical tools, facilitating optimal decision-making and providing a competitive advantage by anticipating trends and strategies. In the present study, we introduce a methodology for the evaluation of classification models based on decision trees. This assessment is conducted through multi-objective optimization, multi-criteria decision-making, and a meticulous analysis of design paradigms. In our approach, three objectives—precision, recall, and number of variables—which at first glance may seem antagonistic, are analyzed simultaneously. We will use machine learning algorithms with diversified structural characteristics. The aim is to elucidate behavioral variations as additional variables are integrated into the system. We argue that rather than focusing solely on conventional metrics in a segmented manner, it is more beneficial to adopt a methodology that synthesizes these objectives.