Analyzing students records to identify patterns of students' performance

• Published in: 2013 International Conference on Research and Innovation in Information Systems (ICRIIS) pp. 544 - 547
• Main Authors: Hoe, Alan Cheah Kah, Ahmad, Mohd Sharifuddin, Tan Chin Hooi, Shanmugam, Mohana, Gunasekaran, Saraswathy Shamini, Cob, Zaihisma Che, Ramasamy, Ammuthavali
• Format: Conference Proceeding Journal Article
• Language: English
• Published: IEEE 01.11.2013
• Subjects: Algorithms; Artificial neural networks; Business; CRISP-DM; Data mining; data modeling clustering; Data models; data preparation; Demographics; Educational institutions; Failure; Mathematical models; Prediction algorithms; Students
• ISSN: 2324-8149; 2324-8157
• DOI: 10.1109/ICRIIS.2013.6716767

Academic failures among university students have been the subject of interest in higher education community. Students drop out due to poor academic performance as early as in the first year of their university enrolment. Many interested parties' debate and try to find reasons for this poor performance. Consequently, the ability to predict a student's performance could be useful in many ways to stakeholders of higher education institutions. This paper discusses the data mining technique used to identify the significant variables that affects and influences the performance of undergraduate students. Students' demographic and past academic performance data are then used to study the academic pattern. Early phases of the CRISP-DM methodology is also described in detail consisting business understanding, data understanding and data preparation. The data modeling and mining tool used identifies the most significant correlation of variables associated with academic success based on the past ten years of demographic and students' performance data of the College of Information Technology, Universiti Tenaga Nasional. Finally, the results from the application of the CHAID algorithm aimed at predicting students' academic success is presented.