An Independent Constructive Multi-class Classification Algorithm for Predicting the Risk Level of Stress Using Multi-modal Data

• Published in: Arabian journal for science and engineering (2011) Vol. 47; no. 8; pp. 10547 - 10562
• Main Authors: Pankajavalli, P. B., Karthick, G. S.
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.08.2022; Springer Nature B.V
• Subjects: Algorithms; Classification; Engineering; Humanities and Social Sciences; Modal data; multidisciplinary; Recommender systems; Research Article-Computer Engineering and Computer Science; Risk levels; Science; Scientific papers; Class imbalance; Stress level; Relative closeness; Macro-averaging; Multi-class classification; Integrated weights; Micro-averaging
• ISSN: 2193-567X; 1319-8025; 2191-4281
• DOI: 10.1007/s13369-022-06643-6

Currently, stress is being a root cause for many health issues, and the necessity of identifying the risk level of stress becomes crucial, which can able to assist the recommender systems by suggesting possible predictive solutions. Most of the decisions made by medical practitioners using the existing classification algorithms are entirely based on the normal and abnormal observations of the patient’s data. The existing algorithms work better for binary classification problems like discriminating the healthy subjects from the stress subjects. There is a mandatory requirement for multi-class classification algorithms which can identify the various risk level of stress. The existing multi-class classification algorithms fail to consider the relative closeness of the features. Also, the classification accuracy is considerably less and suffers from class imbalance issues. In this research paper, an independent constructive multi-class classification (ICMCC) algorithm is proposed to identify and classify the pattern that indicates the risk level of stress based on the relative closeness of features. The study used the synthetic minority over-sampling method to cope up with the class imbalance issue. The experimental study used the multi-modal stress dataset with four classes to evaluate and validate the proposed ICMCC algorithm. The performance of the proposed ICMCC algorithm is compared with the existing state-of-the-art multi-class classification algorithms using the metrics such as accuracy, error rate, precision, recall, and F1-score. From the experimental results, it can be concluded that the proposed algorithm is effective, and the performance is found to be better when compared to the existing multi-class classification algorithms.