Lexicographic optimization-based approaches to learning a representative model for multi-criteria sorting with non-monotonic criteria

Zhen Zhang,Zhuolin Li,Wenyu Yu

2024-09-03

Abstract:Deriving a representative model using value function-based methods from the perspective of preference disaggregation has emerged as a prominent and growing topic in multi-criteria sorting (MCS) problems. A noteworthy observation is that many existing approaches to learning a representative model for MCS problems traditionally assume the monotonicity of criteria, which may not always align with the complexities found in real-world MCS scenarios. Consequently, this paper proposes some approaches to learning a representative model for MCS problems with non-monotonic criteria through the integration of the threshold-based value-driven sorting procedure. To do so, we first define some transformation functions to map the marginal values and category thresholds into a UTA-like functional space. Subsequently, we construct constraint sets to model non-monotonic criteria in MCS problems and develop optimization models to check and rectify the inconsistency of the decision maker's assignment example preference information. By simultaneously considering the complexity and discriminative power of the models, two distinct lexicographic optimization-based approaches are developed to derive a representative model for MCS problems with non-monotonic criteria. Eventually, we offer an illustrative example and conduct comprehensive simulation experiments to elaborate the feasibility and validity of the proposed approaches.

Artificial Intelligence,Machine Learning

What problem does this paper attempt to address?

The paper aims to address the challenge of modeling non-monotonic criteria in Multi-Criteria Sorting (MCS) problems. Specifically, many existing methods typically assume that criteria are monotonic when dealing with MCS problems. However, in practical applications, many MCS problems involve non-monotonic criteria. For example, in the financial sector, the ratio of total liabilities to total assets is often considered a non-monotonic criterion when classifying companies based on their financial status; in medical diagnosis, blood glucose levels also exhibit non-monotonic characteristics.

To tackle this challenge, the paper proposes a lexicographic optimization-based approach to learn representative models that include non-monotonic criteria. The main contributions include:
1. Defining some transformation functions that map marginal values and category thresholds from the original feature space to a UTA-like standard form feature space, ensuring that the transformed marginal value functions and category thresholds maintain the UTA-like standard form and preserve the ranking results of alternatives after transformation.
2. Constructing a set of constraints for non-monotonic criteria in MCS problems and developing a model to check and correct inconsistencies in the preference information provided by decision-makers in allocation examples.
3. Proposing two lexicographic optimization-based methods to learn representative sorting models that include non-monotonic criteria, considering both model complexity and discriminative power, and analyzing the robustness of the proposed methods.

Through these methods, the paper not only addresses the modeling problem of non-monotonic criteria but also enhances the applicability and interpretability of the models, enabling better handling of complex multi-criteria sorting problems in practical applications.