Ensemble Learning Techniques for Classifying Stressed and Unstressed Textual Data
Amit Purushottam Pimpalkar, Devvrat Miglani, Aliya Rizvi, Pranali Dandekar, Sanket Verma, Kushagra Selokar · 2024
In today's digital age, occasional mild stress is commonplace, but excessive stress can significantly affect mental health. Early prediction of stress levels is vital for preventing adverse effects. Automated systems are crucial for accurate predictions, and sentiment analysis, which decodes online conversations, plays a key role. This research focuses on classifying textual data from online conversations into stress and unstressed categories using datasets from X and Reddit. The study aimed to improve sentiment analysis for stress detection in textual data by comparing machine learning approaches. The research utilized NLP techniques and machine learning algorithms to classify stress and non-stress, achieving high accuracy and precision. Employing machine-learning classifiers Multinomial Naive Bayes (MNB), Random Forest (RF), Support Vector Machine (SVM), Decision Tree (DT), and Ensemble Voting Classifier (EVC), the EVC stood out, achieving 87% accuracy for the Twitter dataset and surpassing 92% accuracy on the Reddit dataset, demonstrating its effectiveness in stress classification. It found that ensemble methods, particularly the EVC method, show promise in addressing the complexities of stress detection.