<![CDATA[Sleep apnea is a serious sleep disorder with high global prevalence and is associated with complications such as cardiovascular disease and reduced quality of life, while conventional diagnostic methods remain limited in terms of cost and time efficiency. Therefore, this study aims to classify the risk of sleep apnea and analyze the impact of feature selection on classification model performance. The proposed method employs a quantitative experimental approach using three machine learning algorithms, namely Logistic Regression, Decision Tree, and Random Forest, combined with a Tree Feature Importance-based feature selection technique. The dataset used is the Sleep Health and Lifestyle Dataset obtained from Kaggle, consisting of 374 samples and 13 features related to sleep health and lifestyle factors. The data are split into 80% training and 20% testing sets using stratified sampling and evaluated using accuracy, precision, recall, F1-score, and Area Under the Curve (AUC) metrics. The results indicate that Logistic Regression achieves the most consistent performance with an accuracy of 0.933 and the highest AUC of 0.975 on the Top 10 feature subset, while Random Forest demonstrates strong discriminative capability with an AUC of up to 0.960 on the Top 5 subset. Additionally, Diastolic BP, Systolic BP, BMI Category, and Quality of Sleep are identified as the most influential features. These findings suggest that feature selection effectively preserves model performance while reducing data complexity, leading to a more efficient approach for sleep apnea risk classification. ]]>
Copyrights © 2026
