Machine Learning Model - Ensuring Electronic Exam Quality using Mining Association Rules

Abstract

Overall the objective of the Electronic Exam quality is to help students reach effective learning outcomes. The developed analytical model in this work could be used as an innovative practice to maintain the interrelation structure over the e-exam questions. The objective of this work is to achieve worthy exam questions considering entirely student performance levels. The mining associative rules have been adapted to find the frequent correlations and the causal structures based on the relation of exam answer sets found in the e-exam answer sheets. The dataset consists of 289 electronic exam answer sheets related to e-exam questions of seven courses in the management information system program. All the e-exam answer sheets are available in the Moodle systems at a private institution in the Sultanate of Oman. 28 experiments have been prepared and implemented to validate the practice use of the proposed model in associative rule interest including Support, Confidence, and Lift and Error Rate model respectively. The obtained results show a 10% significant improvement on the exam results subsequently when maintaining the interrelation structure over the e-exam questions. The outcomes of this work are analytical models that ensure the e-exam quality and could act as an advanced warning sign of a negative outcome and will grab the education manager's attention if the e-exam results are slower than the rate preferred.