Comparison of classical machine learning and ensemble techniques in the context of dengue severity prediction
Dengue disease, spread by mosquitoes, affects a large part of the world's population. Early diagnosis is essential to avoid its severe impacts. This paper seeks to compare classical machine learning techniques with ensemble approaches in the early classification of dengue: Dengue without alarm...
- Autores:
-
ARRUBLA HOYOS, WILSON DE JESÚS
Severiche Maury, Zurisaddai de la Cruz
Saeed, Khalid
Gómez Gómez, Jorge Eliecer
De-La-Hoz-Franco, Emiro
- Tipo de recurso:
- Part of book
- Fecha de publicación:
- 2024
- Institución:
- Corporación Universidad de la Costa
- Repositorio:
- REDICUC - Repositorio CUC
- Idioma:
- eng
- OAI Identifier:
- oai:repositorio.cuc.edu.co:11323/13096
- Acceso en línea:
- https://hdl.handle.net/11323/13096
https://repositorio.cuc.edu.co/
- Palabra clave:
- Dengue
Machine learning
Ensemble methods
Classic methods
Staking
Decision tree
- Rights
- embargoedAccess
- License
- Atribución-NoComercial-SinDerivadas 4.0 Internacional (CC BY-NC-ND 4.0)
| Summary: | Dengue disease, spread by mosquitoes, affects a large part of the world's population. Early diagnosis is essential to avoid its severe impacts. This paper seeks to compare classical machine learning techniques with ensemble approaches in the early classification of dengue: Dengue without alarm signs (DNWS), Dengue with alarm signs (DWWS) and Severe Dengue (SD). A dataset available at https://www.datos.gov.co/ from the Colombian government with 53,814 records and 38 attributes is used. The data are processed in Google Colab using Pandas, Matplotlib and Numpy, while Scikit-learn is used for modeling. The results are supported by a detailed confusion matrix, revealing the actual performance of each model and highlighting the superiority of the ensemble approaches over classical techniques in early dengue classification. The Stacking method achieves an accuracy (Acc) of 88%, outperforming all other techniques employed. Among the classical techniques, Decision Tree (DT) achieved the best result with an Acc of 84%. In conclusion, when contrasting the performance of classical techniques and ensemble approaches in early dengue classification, it is highlighted that the latter demonstrate robustness in terms of quality metrics, such as accuracy, recall and F1-Score. This observation suggests that ensemble approaches have the potential to overcome the limitations associated with classical techniq |
|---|