DIGITAL LIBRARY
USING ENSEMBLE DATA MINING APPROACHES TO PREDICTING STUDENT ACADEMIC PERFORMANCE
University of San diego (UNITED STATES)
About this paper:
Appears in: INTED2014 Proceedings
Publication year: 2014
Pages: 4308-4313
ISBN: 978-84-616-8412-0
ISSN: 2340-1079
Conference name: 8th International Technology, Education and Development Conference
Dates: 10-12 March, 2014
Location: Valencia, Spain
Abstract:
Over the recent years, educational data mining has become a very popular research field in the computer science and higher education communities around the world. It develops data mining methods for exploring the data from educational settings and for improving student learning experiences and institutional effectiveness.

Among several educational objectives that institutions of higher education aim to achieve, promoting academic success is a fundamental one. In order to provide an effective learning environment that helps foster student success, we need to understand the important factors that may impact student performance. Furthermore, we need to utilize these factors to build data mining models that accurately identify who are likely academically successful or at-risk. Being able to identify these factors and individual students in each of the two groups with respect to academic success will help institution administrators and faculty offer effective support and intervention services to those who are needed most to succeed.

The paper investigates and applies several ensemble data mining approaches to identify significant factors or attributes that are associated with student academic success and to accurately predict student academic performance for classes from freshman through senior. While the paper focuses on predicting student academic performance through student data collected from a US college, the methodology developed from this work will be applicable to similar or different educational systems. Predicting student academic success is a fundamentally important problem and it has been consistently related to several key areas of higher education administration and planning that include enrollment management, student retention, scholarship and financial aid supervision, and graduation time projection. Finally, the paper demonstrates that ensemble data mining methods can produce superior or competitive models to several other well-known traditional approaches such as decision trees.
Keywords:
Data mining, student academic performance, student retention, enrollment management, graduation time prediction.