Article Preview
Top1. Introduction
Building smart software systems are emerging for medium to large scale systems and in particular when large amount of data need processing. Data processing in many systems needs manual processing by supervisors and administrators of the system. For example, large universities need systems to efficiently process the advising of a large number of students. The manual building of students’ schedules requires a large effort from advisors. The process of determining the courses, students should take in next semester, should be finished within a constrained time. Such process is inefficient when dealing with large number of students. Moreover, advisors usually use a prescribed and general schedule for the students while not considering individual differences among students. In addition, academic advising has been overlooked even though being central for the learning process and the performance of students (Gutiérrez et al., 2019). Most educational organizations provide simple software systems (web interfaces) to support academic advising.
Academic advising is required widely in educational systems to provide direction and assistance to students (Iatrellis et al., 2017). In addition, students are usually allowed to take any courses either from the same major or others (Chang et al., 2016). This kind of choice results in difficulties in selecting the appropriate courses and distracts students from realizing their academic goals. Therefore, many research publications have proposed a dashboard of advising tools to provide access to student achievements and registration (Aguilar et al., 2014). A dashboard can show students’ activities on an electronic management system and provides comparisons of their grades (Fritz, 2011). However, these systems do not provide help in decision making for both the advisor and the student.
There is a need to have a smart system that can automate the decision making to help students in their registration process. The main focus of this research is to build a smart academic advising system that optimizes the registration process and reduces the efforts of the university administration, advisors and students. The current advising system is either a traditional computer-based or paper-based. In addition, such systems consume a considerable amount of time and effort from the department administration and advisors (Biletskiy et al., 2009). Moreover, the process is inaccurate because the available information to the advisor is not always sufficient. The course-load is a main concern for students and it needs to be considered carefully. Several factors may affect the course-load namely, the nature of the subject, student preferences, and other courses taken by the student in the same term. These factors are not considered in the current advising system, therefore, the need for automated, and smart academic advising system is a must. The proposed system will serve the following objectives:
− Helping students register for classes according to their performances. Discovering course registration patterns can help in selecting courses that are supposed to improve students’ performance.
− Drawing associations and dependencies between courses. Finding interdependent courses that were mostly registered together helps the advising process.
− Provide association rules that can help both the advisor and the student in selecting appropriate courses.
To meet these objectives, association rule mining is proposed to find and rank a list of suggested courses that a particular student can consider before registering. Association rules are generated from the history of registration. These rules can direct students to select future courses. The rule X→Y means that a student who has already taken X is most likely to register Y in next semesters. Students usually consult their advisors and preceding students before selecting courses for the next semester. One of the most frequently used data mining techniques in many applications is the association rule mining. It has been used as an underlying technology to improve the decision making process. In this study, the Apriori algorithm is used as a method to build academic advising system.
The rest of the paper is organized as follows. In section 2, the authors discuss related works on academic advising. In section 3, the association rule mining is described. Section 4 discusses the research methodology and the data collection. Section 5 provides a results discussion of the proposed advising system and shows a validation of the association rule mining on 100 students selected from a real data. Section 6 discusses the system limitations. Finally, the authors conclude the work in Section 7.