Strategies For Handling Imbalanced Data In Lightgbm-Based Classification: A Systematic Literature Review
Journal of Social Science (JoSS) · 25.07.2026 · 5 (9) · 782-798
DOI: 10.57185/qfvwrt35 ↗Kurz gefasst
Systematische Übersicht in der Zeitschrift Journal of Social Science (JoSS), online erschienen am 25.07.2026. Verfasst von Adinda Aulia Putri und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).
Abstract (Originalsprache)
Background
Light Gradient Boosting Machine (LightGBM) is used in imbalanced classification, but evidence integrating balancing, performance evaluation, and interpretability in healthcare remains fragmented.
Objective
This review identifies and synthesises LightGBM-based strategies as a foundation for future dengue early detection research.
Methods
A systematic literature review followed the reported PRISMA 2020 selection process using Scopus, ScienceDirect, and PubMed, with 21 articles retained under the original eligibility criteria. Findings were organised thematically.
Results
Four themes emerged: imbalance handling strategies, predictive performance evaluation, applications across domains, and model interpretability. Reported techniques included SMOTE, SMOTE-ENN, SMOTE-Tomek, GAN-based oversampling, and cost-sensitive learning. Studies used different datasets and metrics, limiting direct comparison. The synthesis tables explicitly identify eight distinct primary studies, and complete extraction information for the remaining records is not available in the manuscript. Published performance figures do not establish the independent effects of balancing or explainability without controlled comparisons.
Conclusion
The findings support evaluating LightGBM with minority-class-sensitive metrics, leakage-free validation, and transparent explanations. An integrated Explainable LightGBM framework warrants further testing for dengue detection. These limitations are particularly relevant when evidence is intended to inform high-stakes clinical decisions involving minority-class cases. The review highlights the need for complete study traceability and quality appraisal before drawing stronger comparative conclusions
Zitieren
APA
Putri, A. A., Syahputra, A. I., & Priyatin, B. (2026). Strategies For Handling Imbalanced Data In Lightgbm-Based Classification: A Systematic Literature Review. Journal of Social Science (JoSS), 5(9), 782-798. https://doi.org/10.57185/qfvwrt35
Vancouver
Putri AA, Syahputra AI, Priyatin B. Strategies For Handling Imbalanced Data In Lightgbm-Based Classification: A Systematic Literature Review. Journal of Social Science (JoSS). 2026;5(9):782-798. doi:10.57185/qfvwrt35
BibTeX
@article{Putri2026,
author = {Putri, Adinda Aulia and Syahputra, Arief Ilham and Priyatin, Bilqis},
title = {Strategies For Handling Imbalanced Data In Lightgbm-Based Classification: A Systematic Literature Review},
journal = {Journal of Social Science (JoSS)},
year = {2026},
volume = {5},
number = {9},
pages = {782--798},
doi = {10.57185/qfvwrt35}
}
EUVIMED