Evaluation of Hyperparameter Optimization for Gaussian Naïve Bayes Using a Genetic Algorithm in Liver Disease Classification
Liver disease requires early detection to support appropriate treatment. This study evaluates the effect of Genetic Algorithm (GA)-based hyperparameter optimization on Gaussian Naïve Bayes (GNB) performance in liver disease classification. The dataset was the Indian Liver Patient Dataset (ILPD) from the UCI Machine Learning Repository, comprising 570 samples after preprocessing. Optimization was performed on the var_smoothing parameter within the range of to , using accuracy as the fitness function. The GA used a population of 10 individuals and five generations, while model evaluation employed 10-fold cross-validation. The baseline GNB achieved mean accuracy of 56.14%, precision of 96.60%, recall of 39.93%, and F1-score of 56.04%. After GA optimization, mean accuracy increased to 56.32%, precision remained at 96.60%, recall reached 40.18%, and F1-score reached 56.31%. These improvements were small, while the high number of false negatives indicated limitations in identifying liver disease samples. The results show that GA did not improve GNB performance substantially under the evaluated configuration and search range.
Authors
- Taghfirul Azhima Yoga Siswa (ORCID: https://orcid.org/0000-0003-2017-8538)
- Muhammad Ridwan
- Nurjanna Nurjanna
Institutions
- Universitas Muhammadiyah Kalimantan Timur (ID)
Publication Details
- Journal
- JOINS (Journal of Information System)
- Published
- 2026-11-12
- DOI
- https://doi.org/10.33633/joins.v11i2.17705
- Primary Topic
- Artificial Intelligence in Healthcare
- Type
- article
- Field-Weighted Citation Impact
- 0.00