Banca de DEFESA: VINICIUS DE OLIVEIRA WATANABE

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : VINICIUS DE OLIVEIRA WATANABE
DATE: 02/10/2025
TIME: 14:00
LOCAL: A defesa será realizada por videoconferência
TITLE:

Fairness Score: a composite metric for fairness and performance evaluation

KEY WORDS:

algorithmic fairness; credit risk; fairness metrics; imbalanced data; machine learning.

PAGES: 30
BIG AREA: Ciências Sociais Aplicadas
AREA: Economia
SUMMARY:

The rapid adoption of machine learning in critical domains has intensified concerns about algorithmic bias and the fairness–performance trade-off. While existing literature often focuses on mitigation techniques, evaluating models typically involves assessing performance and fairness in isolation. This study proposes a Fairness Score, a composite metric that unifies predictive performance (F1-score) and fairness metrics (Demographic Parity Ratio, Equalized Odds Ratio, Predictive Rate Parity) into a single evaluative framework. A weighting parameter, alpha, allows decision-makers to adjust the relative importance assigned to accuracy and fairness according to contextual priorities. We evaluate the proposed score in a credit risk prediction context, using an imbalanced dataset from a Brazilian microcredit program. Results indicate that the Fairness Score provides a robust and balanced assessment, effectively capturing the trade-offs between fairness and performance across different models and demographic scenarios. The score offers a practical and transparent tool for developing more responsible and reliable automated decision-making systems.

COMMITTEE MEMBERS:
Presidente - 1642911 - DANIEL OLIVEIRA CAJUEIRO
Externo ao Programa - 1997721 - HERBERT KIMURA - nullInterno - 1550794 - JOSE GUILHERME DE LARA RESENDE
Externo à Instituição - ROBERT ALDO IQUIAPAZA COAGUILA - UFMG