Abstract:
Outcome Based Education focuses on the performance of students in term of skills, knowledge and capability over 4 year degree plan. Unfortunately to achieve the intended learning outcomes of a course, students always seems to have trouble in term of choosing course which brings low performance measure for them. In educational environment, recommendation system is an intelligent agent which recommends possible alternatives to students keeping in view his previous academic records or other personal information. The educational information system consist of enormous amount of potential data which can be used to improve educational outcomes using specific attribute set that would requires different analysis of mining the problem.
In this research, we have proposed the integration of Education Data Mining environment with respect to Outcome Based Education through two modules. In first module, we have implemented J48 decision tree algorithm as data mining technique to explore influential mechanism of recommendation system. It groups students’ academic result of core courses into three groups; computing core, software engineering core and supporting core where a student would be recommended course group on the basis of the performance against it. It further recommends elective courses mapped against the recommended core course group. In second module, students’ academic result are classified into the assessment measure of CLO against courses. This assessment measure would help to ensure accurate assessment level achieved by students to recommend courses that would help them to accomplish the failed PLO. To validate the course group recommendation model, we compared the J48 algorithm with other binary classification techniques such as Random forest, Naïve Bayes, SMO, Multi-layer perceptron and Logistics by using the sample educational data. The results showed that the proposed algorithm had the best accuracy in term of predicting results against the average time taken by the dataset.
