GRSFA: Recommending Course Grade for Improving Academic Performance of Students using Filtered Associator Algorithm in Education Sector

Recommendation system acts as a information filtering system that provide the suggestions to users based on many different factors. In the current study, a framework for recommendation system called CGRSFA is developed for recommending the grade of course. For this framework, semester-wise, year-wise and overall grade information of students’ courses is stored in ten datasets. This framework uses real data gathered from university site related to Four Year Bachelor of Technology - Computer Science and Engineering Programme, counting on overall 11510 instances and 42 courses. Relevant rules which indicates courses dependencies and courses prerequisites for other courses are found using this framework for each dataset e.g. the meaning of the rule “System_Programming=A + → Compiler_Construction=A+” is that if student receives ‘A’ grade in System Programming course then that student will received ‘A’ grade in Compiler Construction course also as System Programming course is the prerequisite to the course Compiler Construction. Rules which are not relevant are irrelevant rules and such rules are discarded. Result of four association rule algorithms such as Apriori Association Rule, Filtered Association algorithm, Predictive Apriori Association Rule and Tertius Association Rule algorithms are compared based on relevant and irrelevant rules for selecting the best association rule algorithm for the grade recommendation system. The algorithm Filtered Associator algorithm is selected among these algorithm for the grade recommendation system. Filtered Associator algorithm is used to find the correlation among the courses. In Filtered Associator algorithm, first the grade dataset is filtered using the filtering method - Reservoir sampling Algorithm to remove the data items from dataset that do not meet certain criteria and then Apriori association rule algorithm is applied on the filtered dataset. Association rules generated along with the support parameter value for one of the dataset D6 is given and explained in the current study. If the support parameter value of obtained association rules is increased then the most optimal association rules for maximum support value are generated using Filtered Associator algorithm for the grade recommendation system. Number of association rules generated for remaining nine datasets along with support parameter value is also presented in the experimental result. This recommendation system is useful for instructor as well as students for improving academic performance. This system can also be used in MOOCs for recommending the course grade.