Modelling an Innovative Machine Learning Model for Student Stress Forecasting

Abstract

Student stress has become an important distress in an educational environment, affecting academic presentation and overall well-being. Traditional methods of stress prediction in students lack accuracy and fail to adapt to the dynamic nature of stress factors. This study aims to develop and validate an advanced machine-learning model for predicting student stress. Data was collected from a various group of students through studies and behavioural valuations, capturing numerous stress-related indicators. Data pre-processing for scaling the features of the model was achieved through min-max normalization. Study utilized a Fine-tuned Squirrel Search-driven Light Gradient Boosting Machine (FSS-LGBM) model, which integrates advanced optimization techniques with Light GBM for enhanced predictive performance. The FSS-LGBM model validated superior performance associated with convolutional machine learning models, achieving higher accuracy (98%), recall (99%), precision (97%), and F1-score (97%) in student stress prediction. The advanced machine learning model developed in this study, incorporating FSS-LGBM, offers a promising approach for predicting student stress. By employing advanced data pre-processing and optimization techniques, the model offers a proactive tool for early stress detection, potentially leading to enhanced mental well-being intrusions and support for students.