IBEDO-DE: A Novel ChatGPT-Enhanced Model for ImprovingEducational Outcomes through Data-Driven Insights and Student Perceptions

Abstract The rapid advancement in educational technologies has ushered in a transformative era in higher education, necessitating rigorous research to harness these innovations for enhancing educational quality and performance. Understanding students’ perceptions of emerging technologies like ChatGPT is critical for tailoring educational tools that meet their learning needs and preferences. Objective: The research is focused on analyzing students' perceptions of ChatGPT in an educational setting. It utilizes Learning Analytics (LA) and Feature Selection (FS) techniques on a dataset initially gathered from an online questionnaire regarding student opinions on ChatGPT. The study seeks to uncover key insights into the effectiveness and reception of ChatGPT services in higher education. Methods: This study suggests the Exponential Distribution Model - Differential Evolution (EDO-DE) Algorithm to address the FS problem. The EDO-DE chooses the considerable useful and related attributes. In addition, an improved EDO-DE (IBEDO-DE) is proposed, which improves searching and exploitation abilities by integrating the local search (LS) strategy and periodic mode boundary handling technique. This assists in reducing dimensionality and improving the accuracy of classification. Two commonly employed ML models, support vector machine and k-nearest neighbor, were utilized as performance evaluators to assess the effectiveness of chosen attributes. The binary versions of sixteen recent optimize were compared and analyzed. The objective is to identify critical features influencing students’ perceptions of ChatGPT services. Results: The analysis is expected to reveal significant features that determine the utility and acceptance of ChatGPT services among students. The suggested IBEDO-DE surpasses by obtaining an accuracy of 89.9 in certain real datasets while reducing the feature set by up to 0.5. It has been statistically proven that the suggested IBEDO-DE is highly competitive based on the Wilcoxon rank sum test (with alpha=0.05). Conclusion: Understanding students' perceptions of ChatGPT through advanced LA and FS techniques provides valuable insights for educators and technologists. By identifying critical features that enhance the learning experience, higher education institutions can better align technological tools with educational goals, achieving higher educational efficacy and student satisfaction.