Heliyon (Jun 2024)
Development of an AI predictive model to categorize and predict online learning behaviors of students in Thailand
Abstract
This qualitative study has three objectives: (1) to develop a predictive AI model to categorize the online learning behavior of Thai students who study through a Thai Massive Open Online Course (MOOC); (2) to categorize students’ online behavior in a Thai MOOC; and (3) to evaluate the prediction accuracy of the developed predictive AI models. Data were collected from 8000 learners enrolled in the KMUTT015 course on the Thai MOOC platform. The k-means clustering algorithm classified learners enrolled in the Thai MOOC platform based on their online learning behaviors. The decision tree algorithm was used to assess the accuracy of the AI model prediction capability. The study finds the predictive AI model successfully categorizes learners based on their learning behaviors and predicts their future online learning behaviors in the online learning environment. The k-means clustering algorithm yields three groups of learners in the Thai MOOC platform: High Active Participants (HAP), Medium Active Participants (MAP), and Lurking participants. The findings also indicate high predictive accuracy rates for each behavioral group (HAP cluster = 0.98475, Lurking participants cluster = 0.967625, and MAP cluster = 0.955375), indicating the proficiency of the AI predictive model in forecasting learner behavior. The results of this study will benefit the design of online courses that respond to the needs of students with different online learning characteristics and help them achieve a high level of academic performance.
