Komparasi Algoritma Machine Learning dengan Hyperparameter Tuning pada Prediksi Level Obesitas

Abstract

Obesity has become a growing global health concern due to its association with various chronic diseases, including diabetes mellitus, hypertension, and cardiovascular disorders. This study investigates the performance of Support Vector Machine (SVM), Random Forest, and Gradient Boosting algorithms enhanced through hyperparameter tuning for obesity level prediction. The study utilized the Obesity Level dataset, which contains 2,111 instances, 16 input features, and one target class. Following the data preprocessing stage, 2,087 instances were included in the analysis. Correlation analysis was conducted to identify the nine most relevant features used for model development. Experimental results demonstrate that the Random Forest model achieved the best predictive performance compared to SVM and Gradient Boosting, with an accuracy of 90.91%, precision of 91.19%, recall of 90.91%, and an F1-score of 90.94%. The dataset was partitioned into training and testing subsets using an 80:20 split. Furthermore, the optimal model was deployed in a Streamlit-based web application to facilitate obesity level prediction. These findings suggest that Random Forest, when optimized through hyperparameter tuning, provides a reliable and effective approach for multiclass obesity classification.