An investigation of novel features for predicting student happiness in hybrid learning platforms – An exploration using experiments on trace data

Abstract

The current study investigated the novel factors influencing informatics students' happiness and satisfaction using statistical and machine learning algorithms. A statistical t–test proved that the students in the younger age group (=21) who had enrolled in master's programs (M.Sc.) that hybrid learning was a secure mode of education during the pandemic (p < 0.05). The happiness of students has had a significant influence on the challenges and prospects for online education that involve safety (p < 0.05). It was determined that factors such as time and cost utilization with management, university initiatives and assistance, and challenging group tasks substantially affected student satisfaction (p < 0.05). The paper's findings recommended some vital issues such as Internet disconnection during the lesson, Less support during technical glitches or errors, Less competitive environment, Possibility of cheating in exams due to lack of surveillance, Less focused and interactive during hybrid learning, important assistance and support, and safety modes for higher educational institutions during a pandemic. Using a maximum accuracy of 88%, Random Forest (RF) outperformed Logistic Regression (LR), Xtream Gradient Boosting (XGBoost), and other classifiers to predict student happiness based on hybrid learning features. The RF algorithm's superior performance was validated and confirmed by all performance metrics, such as F1-score, precision, recall, and specificity.

Keywords

• Factor analysis
• Hybrid learning
• Prediction
• Classification
• Satisfaction
• Happiness