<![CDATA[Sleep disorders are serious health conditions that are closely associated with an increased risk of hypertension and cardiovascular disease. Considering the high cost of conventional clinical diagnostic procedures, data mining techniques for early detection offer an effective alternative. This study aims to conduct a comparative performance analysis between the Decision Tree (C4.5) and Random Forest algorithms in classifying sleep disorders (None, Insomnia, and Sleep Apnea). To address class imbalance issues, this research utilizes the secondary Sleep Health and Lifestyle dataset. To evaluate model stability, a rigorous validation method, Stratified 10-Fold Cross-Validation, is employed. The experimental results indicate that the Random Forest algorithm outperforms the Decision Tree, achieving an average accuracy of 91.41% and a Kappa value of 0.8473. The primary advantage of the Random Forest algorithm lies in its ability to significantly improve the detection sensitivity of the Insomnia class to 88.31%. Based on feature importance analysis, diastolic blood pressure and the BMI category (Overweight) are identified as the most influential features in the diagnostic process. Random Forest is therefore considered a more accurate and stable model for medical decision support systems.]]>
Copyrights © 2025
