P. Huidobro, A. Bouchet, S. Díaz Vázquez, S. Montes
Penalty functions have proven to be a powerful tool for numerical aggregation; however, their adaptation to interval-valued data remains an open challenge. In this paper, we propose a new framework for penalty functions specifically designed for interval data. Our approach defines penalty functions based on coherent and lower semi-continuous distances between intervals, ensuring their compatibility with compact domains. By combining these distances with appropriate aggregation functions, the proposed methodology guarantees the existence of minimizers, providing a robust foundation for interval-based optimization. We demonstrate the practical applicability of these penalty functions by using linguistic data from human evaluations to select representative values from interval datasets, showcasing how they enhance the interpretability and accuracy of interval-based decision-making.
Keywords: Penalty functions, Interval-valued data, Aggregation functions, Interval-based optimization
Scheduled
Data Analysis in Social Sciences
June 10, 2025 7:00 PM
Auditorio 2. Leandre Cristòfol