A0258
Title: Exploring metric performance for binary classification in unbalanced data: A comparative study
Authors: Jorge Luis Bazan - University of Sao Paulo (Brazil) [presenting]
Alex de la Cruz Huayanay - Pontificia Universidad Catolica del Peru (Peru)
Abstract: Addressing the recurring challenge in statistical modeling, binary classification, becomes even more complex when encountering imbalanced data in response categories. Extensive literature presents various models and algorithms for binary classification, each leveraging explanatory variables differently. To assess their effectiveness, numerous computing methods have been proposed, utilizing predictive measures to evaluate model performance. The focus is on the performance analysis of prominent metrics found in literature, and its application is extended to the realm of econometrics. Through the investigation, the aim is to provide insights into the efficacy of these metrics and their applicability in real-world scenarios.
