Stress Detection using Naive Bayes and Decision Tree Classifiers

Abstract

A large percentage of people deal with the ubiquitous problem of stress in our ever-busier lives, which includes both emotional and physical strains. Regrettably, a lack of awareness causes many people to fail to identify the symptoms of stress. Early detection of stress is essential because, if left unchecked, its effects can lead to serious health issues that progressively erode one's physical and mental health. Automated methods help to lower the risks related to stress that go untreated in addition to making stress detection easier. Although a number of models, including the multi-attention model, the Factor Graph model, and the Personal Knowledge model, have been developed for social media blogs, their efficacy has been limited because they have mainly examined single-line text. To overcome this constraint, a Decision Tree and Naive Bayes classifiers has been presented, with the goal of identifying stress from multi-line text data in the dataset that includes user-generated content that is both stressed and unstressed. The method has practical potential as evidenced by the promising accuracy of the experimental results.