Modeling Sleep Health and Lifestyle Using Supervised Learning Algorithms

Abstract

With the realization of the importance of sleep quality as an indicator of general well-being, this work uses the strength of machine learning to discover significant trends in information about lifestyles and health to make predictions regarding sleep health. Based on the Sleep Health and Lifestyle Dataset that consists of 373 instances (rows) and 13 features (columns), including demographic, physiological, and lifestyle-related data, one can classify the target variable Quality of Sleep as a categorical attribute, defining the task in terms of classification. Different machine learning models were implemented and compared by means of precision, recall, accuracy, and F1 score: Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Random Forest, Logistic Regression, Decision Tree, Gradient Boosting, Naive Bayes, and AdaBoost. Of these, the models of the ensemble type were better performers, and the Random Forest model produced the best outcomes in all measures: 98.67% accuracy, 98.74% precision, 98.67% recall, and 98.66% F1 score. The other schemes, Decision Tree and Gradient Boosting, also performed well, and SVM received the lowest scores. The results emphasize the usefulness of ensemble methods in modeling complicated and non-linear relationships in multifactor health data. The results reinforce the possibilities of machine learning in relation to data-driven, individualistic health-related recommendations and early interventions during sleep health management.