11/01/2024
By Zakkiyya Witherspoon

The School of Education invites you to attend a doctoral dissertation defense by Tonghui Xu “Applying the Knowledge Discovery in Database Process Model as a Framework to Enhance Educational Data Mining and Learning Analytics”

Candidate: Tonghui Xu
Degree: Doctoral in Research & Evaluation in Education
Defense Date: Thursday, November 14, 2024
Time: 10 a.m.
Location: Virtual, Zoom link
Thesis/Dissertation Title: "Applying the Knowledge Discovery in Database Process Model as a Framework to Enhance Educational Data Mining and Learning Analytics”

Dissertation Committee
Hsien-Yuan Hsu (Co-chair), Ph.D., Associate Professor, School of Education, College of Fine Arts, Humanities and Social Sciences, University of Massachusetts Lowell
Yan Wang (Co-chair), Ph.D., Associate Professor, Department of Psychology, College of Fine Arts, Humanities and Social Sciences, University of Massachusetts Lowell
Xiaobai (Bob) Li, Ph.D., Professor, Department of Operations and Information Systems, Manning School of Business, University of Massachusetts Lowell
David J. Willis, Ph.D., Associate Professor, Department of Mechanical and Industrial Engineering, Francis College of Engineering, University of Massachusetts Lowell


Abstract
With the introduction of computer technologies into educational fields, a substantial amount and variety of educational big data are being generated across different educational environments, especially in higher education institutions (HEIs). Due to these changes, many educational researchers have begun to acknowledge the significance of utilizing big data in education. Educational big data has significantly advanced educational research, especially in the fields of educational data mining (EDM) and learning analytics (LA), both of which are rapidly expanding research areas within the realm of educational big data. However, there are also challenges that prevent educational researchers from effectively conducting studies in this field. However, EDM/LA studies also present challenges for educational researchers, such as a lack of familiarity with the complexity of machine learning/data mining models, difficulties in pre-processing educational big data, and issues related to the design of computer-based education. This challenge also results in the issue of information overload when educators use big educational data for EDM/LA studies.
Many researchers are beginning to explore ways to address the challenges faced in EDM/LA studies. The literature indicates that the knowledge discovery in databases (KDD) process model can serve as a guiding framework to assist educators in designing studies involving big educational data for EDM/LA, thereby mitigating the impact of information overload. However, the KDD process model was not initially designed specifically for educational research. To address this gap, Romero and Ventura (2020) proposed the educational data mining and learning analytics knowledge discovery (EDM/LA KD) cycle process model for conducting studies on big educational data. Despite this, many educational researchers remain unfamiliar with the KDD model.
This study has three main objectives. Firstly, it seeks to familiarize educational researchers with the EDM/LA KD cycle process model. Secondly, it aims to demonstrate how to use the KDD process model to reduce information overload when educational researchers conduct EDM/LA studies in HEIs using various types of educational big datasets. This will be achieved through three articles: (a) computer-based education in LA study, (b) machine learning/data mining models in EDM study, and (c) integrated machine learning/data mining and educational statistical models in EDM study. Finally, this study discusses how the EDM/LA KD cycle process model can aid educational researchers in crafting big data studies within HEIs and mitigating the impact of information overload.
Overall, the findings suggest that the KDD process model can support educational researchers in conducting EDM and LA studies, including research design, data pre-processing, analysis, model evaluation, interpretation, and the discovery of new knowledge in education while mitigating information overload. By guiding researchers through systematic steps in data mining and analytics, the model enhances the effectiveness of applying EDM and LA to educational big datasets.