Prediction of University Students' Subjective Well-Being with Sleep and Physical Activity Data using Classification Algorithms

Abstract

Daily activities affect mental health. One of the most used scales is "subjective well-being (SWB)", which is a self-reported questionnaire. This study aimed to predict SWBs using step count, heart rate and sleep duration data from sensors instead of questionnaires. NetHealth data from the University of Notre Dame1 has been used. Attributes included average daily steps, average heart rate, heartbeat standard deviation, average sleep duration, and sleep duration deviation. Preprocessing, processing, classification, and evaluation followed. Naive Bayes, K-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Ensemble classifiers were used. Performance metrics include accuracy, precision, recall, F1-Score, and ROC (Receiver Operating Characteristic) curves. Model accuracy was 62%. This indicates that machine learning could be beneficial in detecting SWB levels using sensor data. © 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the 26th International Conference on Knowledge-Based and Intelligent Information & Engineering Systems (KES 2022).