Modeling Filipino academic affect during online examination using machine learning

Abstract

A learning environment is comprised of emotional and cognitive activities. The need to incorporate the measurement of student affect during the learning assessment is necessary. This study measured academic affect during assessment and developed an online examination multi-modal academic affect model. Using the matic analysis and expert opinion, the study generated the presence of the following academic affect: relaxed, curious, bored, frustrated, and distracted. The academic affects were annotated with a computed interrater kappa value of 0.62. Annotation by the coders was mapped with the features extracted from a behavior analysis toolkit that produced the Filipino Online Examination Multi-Modal Affect Dataset. The model yielded an accuracy of 92.66%.