Predicting Psychological State of Students based on Physiological, Behavioral, and Environmental Attributes

Comprehension of psychological states is important because it can lead to better mental health support, improved educational outcomes, and enhanced workplace productivity. By understanding an individual's emotional and cognitive condition, timely interventions can be made to reduce stress, anxiety, and other negative psychological states. This research focuses on predicting psychological states using a diverse dataset containing physiological, environmental, behavioral, psychological, and demographic features. In this study physiological indicators such as heart rate variability, skin response, and brain activity were captured alongside environmental factors like ambient noise and light levels. Behavioral features such as focus duration and typing speed were included and analyzed in relation to cognitive load and emotional states. The data was preprocessed, converting categorical strings into numerical values for compatibility with machine learning algorithms. IBM Watson selected the Decision Tree Classifier and Snap Decision Tree Classifier for modeling. Performance analysis revealed the most accurate model in this study was the Decision Tree Classifier, with an accuracy of about 30%. While the results were not ideal, possible methods for improving the accuracy are given in the future work section. In general, the work in this paper analyzes the potential for identifying the correlation between behavioral, physical, and environmental attributes and one's psychological state.