Student Performance Monitoring in an Online Environment

Abstract

Nowadays, web-based educational systems are being installed more and more by universities, schools, businesses, and even individual instructors in order to add web technology to their courses and to supplement traditional face-to-face courses. Therefore identifying factors affecting the success and failure of e-learning has become essential. On the other hand systems used to demonstrate an e-learning environment accumulate a vast amount of data which is very valuable for analyzing the content of the courses and their usage from the learners has led to the deployment of data mining. Therefore the purpose of this research is to identify factors affecting the success and failure of e-learning student. This thesis describes different data mining technologies used to identify performance factors and provide the most significant factors correlated to the success of a student in a Virtual Learning Environment. First an initial analysis of data set was done to understand the data set and based on the nature of the dataset it was decided to classify students based on their final results of Pass and Fail in the final exam which is at the end of the module. Multilayer Perception to identify factors that highly affect to the final result, CHAID Decision Tree to identify how distribution of factors affected and the relationship and direction among the factors for the final result and Binary Logistic Model to identify the probability of a student on final result and also for designing, are used as data mining techniques. Generated models are evaluated to identify the best fitting model for the dataset using classification tables, Gain charts, Lift charts, ROC curves and area under curve. The research shows that Quizzes, Assignments and interactive media are the most influence factors while demographical factors like highest education, disability, gender and age are not to bother for a better final result. Among influenced factors number of interactions on quizzes have major impact on final result of a student. The probability of having a pass can be gain for a student can be gain in periodically and see in which factors he/she should focus on. The results and models of this study also can be used to monitor the performance of the student by giving advices in advance in order to gain better result at the end. Also designers can design their learning materials focusing mainly on enhancing the performance of the student. Key Words: Virtual Learning Environment, data mining, monitoring