EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-SA 4.0

Evaluación del Rendimiento de Modelos de Machine Learning para el Análisis de Sentimientos en Tweets en Español bajo Diferentes Escenarios de Desbalance de Clases

Roly Steeven Cedeño Menéndez, José Alberto León Alarcón, Jandry Hernando Franco Cantos

Latin-American Journal of Computing · 07.07.2026 · 13 (2) · 30-42

DOI: 10.33333/lajc.vol13n2.02 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Latin-American Journal of Computing, online erschienen am 07.07.2026. Verfasst von Roly Steeven Cedeño Menéndez und 2 weiteren Autor:innen. Die Arbeit nennt 33 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-SA 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Sentiment analysis has gained significant relevance in the classification of polarities in unstructured texts; however, one of its main challenges is class imbalance, which significantly affects the performance of machine learning models. Therefore, this study compares the performance of six classification algorithms (Naive Bayes, SVM, Logistic Regression, Decision Tree, Random Forest, and XGBoost) on Spanish-language tweets, considering three scenarios: balanced distribution, moderately imbalanced, and highly imbalanced. The evaluation was conducted using the metrics of accuracy, precision, recall, and f1-score. From this, it is evident that linear models achieve better performance in balanced scenarios, although their effectiveness decreases as the imbalance increases. Meanwhile, Naive Bayes maintains a more stable behavior across scenarios, and XGBoost represents a competitive alternative. Additionally, it is observed that the impact of imbalance is not adequately reflected by accuracy, making the f1-score a more representative metric. Altogether, these findings highlight the importance of considering class imbalance in model selection.

Zitieren

APA

Menéndez, R. S. C., Alarcón, J. A. L., & Cantos, J. H. F. (2026). Evaluación del Rendimiento de Modelos de Machine Learning para el Análisis de Sentimientos en Tweets en Español bajo Diferentes Escenarios de Desbalance de Clases. Latin-American Journal of Computing, 13(2), 30-42. https://doi.org/10.33333/lajc.vol13n2.02

Vancouver

Menéndez RSC, Alarcón JAL, Cantos JHF. Evaluación del Rendimiento de Modelos de Machine Learning para el Análisis de Sentimientos en Tweets en Español bajo Diferentes Escenarios de Desbalance de Clases. Latin-American Journal of Computing. 2026;13(2):30-42. doi:10.33333/lajc.vol13n2.02

BibTeX
@article{Menndez2026,
  author = {Menéndez, Roly Steeven Cedeño and Alarcón, José Alberto León and Cantos, Jandry Hernando Franco},
  title = {Evaluación del Rendimiento de Modelos de Machine Learning para el Análisis de Sentimientos en Tweets en Español bajo Diferentes Escenarios de Desbalance de Clases},
  journal = {Latin-American Journal of Computing},
  year = {2026},
  volume = {13},
  number = {2},
  pages = {30--42},
  doi = {10.33333/lajc.vol13n2.02}
}
Kostenlos starten →