Evaluation of Feature Selection and Multi-Class Prediction Methods For Metal Stress

Abstract

Mental stress is a major issue in modern society, especially among young people. The pressure is on the age group that was formerly considered the most carefree. The modern epidemic of stress is a major contributor to many health problems, including anxiety, insomnia, eating disorders, and even death. Sensory equipment including wearable sensors, electrocardiogram (ECG), electroencephalogram (EEG), and photo plethysmography (PPG), as well as varied situations like driving, studying, and working, inform the stress detection techniques employed. This study uses Electrocardiogram (ECG) to examine how stress detection methods vary across contexts including driving, studying, and working. Voting classifier performance is improved by attempting to identify the optimal feature set through the Correction feature selection approach. State-of-the-art results are achieved by applying the proposed methods to the benchmark SWELL-KW dataset.