Performance of Classification Algorithms Under Class Imbalance: Simulation and Real-World Evidence
Class imbalance is a persistent challenge in machine learning, particularly in high-stakes applications such as medical diagnostics, bioinformatics, and fraud detection, where the minority class often represents critical cases. While prior research has examined the effect of imbalance on classifier...
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Artículo |
| Lenguaje: | inglés |
| Publicado: |
2025
|
| Acceso en línea: | Enlace permanente (DOI o Handle) Enlace permanente (DOI o Handle) |
| Etiquetas: |
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1881508987865661440 |
|---|---|
| author | Arshad, Iqra Umair, Muhammad Jan, Faheem Iftikhar, Hasnain Canas Rodrigues, Paulo Ivan Gonzales Medina, Ronny López Gonzales, Javier Linkolk |
| author_browse | Arshad, Iqra Canas Rodrigues, Paulo Iftikhar, Hasnain Ivan Gonzales Medina, Ronny Jan, Faheem López Gonzales, Javier Linkolk Umair, Muhammad |
| author_facet | Arshad, Iqra Umair, Muhammad Jan, Faheem Iftikhar, Hasnain Canas Rodrigues, Paulo Ivan Gonzales Medina, Ronny López Gonzales, Javier Linkolk |
| author_sort | Arshad, Iqra |
| collection | Producción científica |
| description | Class imbalance is a persistent challenge in machine learning, particularly in high-stakes applications such as medical diagnostics, bioinformatics, and fraud detection, where the minority class often represents critical cases. While prior research has examined the effect of imbalance on classifier performance, little attention has been paid to establishing practical guidelines for the minimum proportion of minority samples required to achieve reliable sensitivity. In this study, we conduct extensive simulations using synthetic datasets and evaluate five widely used classification algorithms: Logistic Regression (Logit), Support Vector Machines (SVM), Random Forest, XGBoost, and Neural Networks (NNs). Our analysis reveals that logistic regression is more effective in identifying minority-class instances under an imbalanced class distribution in terms of F1 score and sensitivity, whereas Neural Network slightly performs better for a balanced-class distribution than logistic regression. Importantly, we identify a practical threshold for minority class representation: classifier sensitivity declines sharply when positive samples fall below approximately 25–30%. This finding is validated on eight real-world datasets, including large-scale applications, where Neural Networks and XGBoost demonstrate superior sensitivity. By establishing an actionable threshold, this study contributes practical guidance for dataset design and model selection in imbalanced classification problems. |
| format | Article |
| id | oai:cris.upeu.edu.pe:20.500.15037/407 |
| institution | UPeU |
| language | English |
| publishDate | 2025 |
| publishDateRange | 2025 |
| publishDateSort | 2025 |
| record_format | dspace |
| spelling | oai:cris.upeu.edu.pe:20.500.15037/4072026-11-02T22:00:04ZPerformance of Classification Algorithms Under Class Imbalance: Simulation and Real-World EvidenceArshad, IqraUmair, MuhammadJan, FaheemIftikhar, HasnainCanas Rodrigues, PauloIvan Gonzales Medina, RonnyLópez Gonzales, Javier LinkolkClass imbalance is a persistent challenge in machine learning, particularly in high-stakes applications such as medical diagnostics, bioinformatics, and fraud detection, where the minority class often represents critical cases. While prior research has examined the effect of imbalance on classifier performance, little attention has been paid to establishing practical guidelines for the minimum proportion of minority samples required to achieve reliable sensitivity. In this study, we conduct extensive simulations using synthetic datasets and evaluate five widely used classification algorithms: Logistic Regression (Logit), Support Vector Machines (SVM), Random Forest, XGBoost, and Neural Networks (NNs). Our analysis reveals that logistic regression is more effective in identifying minority-class instances under an imbalanced class distribution in terms of F1 score and sensitivity, whereas Neural Network slightly performs better for a balanced-class distribution than logistic regression. Importantly, we identify a practical threshold for minority class representation: classifier sensitivity declines sharply when positive samples fall below approximately 25–30%. This finding is validated on eight real-world datasets, including large-scale applications, where Neural Networks and XGBoost demonstrate superior sensitivity. By establishing an actionable threshold, this study contributes practical guidance for dataset design and model selection in imbalanced classification problems.Articlehttps://doi.org/10.1109/access.2025.362026410.1109/access.2025.3620264https://hdl.handle.net/20.500.15037/407W4415048073enIEEE Access2169-35362025-01-012026-06-13T08:12:53Z2026-06-13T08:12:53Z |
| spellingShingle | Arshad, Iqra Umair, Muhammad Jan, Faheem Iftikhar, Hasnain Canas Rodrigues, Paulo Ivan Gonzales Medina, Ronny López Gonzales, Javier Linkolk Performance of Classification Algorithms Under Class Imbalance: Simulation and Real-World Evidence |
| title | Performance of Classification Algorithms Under Class Imbalance: Simulation and Real-World Evidence |
| title_full | Performance of Classification Algorithms Under Class Imbalance: Simulation and Real-World Evidence |
| title_fullStr | Performance of Classification Algorithms Under Class Imbalance: Simulation and Real-World Evidence |
| title_full_unstemmed | Performance of Classification Algorithms Under Class Imbalance: Simulation and Real-World Evidence |
| title_short | Performance of Classification Algorithms Under Class Imbalance: Simulation and Real-World Evidence |
| title_sort | performance of classification algorithms under class imbalance simulation and real world evidence |
| url | https://doi.org/10.1109/access.2025.3620264 https://hdl.handle.net/20.500.15037/407 |
| work_keys_str_mv | AT arshadiqra performanceofclassificationalgorithmsunderclassimbalancesimulationandrealworldevidence AT umairmuhammad performanceofclassificationalgorithmsunderclassimbalancesimulationandrealworldevidence AT janfaheem performanceofclassificationalgorithmsunderclassimbalancesimulationandrealworldevidence AT iftikharhasnain performanceofclassificationalgorithmsunderclassimbalancesimulationandrealworldevidence AT canasrodriguespaulo performanceofclassificationalgorithmsunderclassimbalancesimulationandrealworldevidence AT ivangonzalesmedinaronny performanceofclassificationalgorithmsunderclassimbalancesimulationandrealworldevidence AT lopezgonzalesjavierlinkolk performanceofclassificationalgorithmsunderclassimbalancesimulationandrealworldevidence |