Modelling Course Difficulty Indexes to Enhance Students Performance and Course Study Plans

Abstract

The overall aim of this thesis has been to address the task of Long-Term Course Planning (LTCP), advising students to choose the learning plan that suits them best, and reducing the drop-out rate. This overall objective is broken down into several sub-objectives as described: - To propose a sequential pattern mining approach that analyses, in a descriptive way, different study plans for similar students. This study plan considers historical data of students with good grades in terms of courses and average grades of the degree. - To present a course difficulty index to measure the eligibility of a specific course. A maximum difficulty value is considered for students to make choices according to the metric provided. - To propose a web application to make it possible to obtain the value of the difficulty index for different subjects of different degrees. - To apply the above sub-objectives to a real problem with data from King Abdulaziz University (KAU), one of the most important universities in Saudi Arabia, located in Jeddah. The analyzed data came from multiple sources from heterogeneous environments such as the ODUSPLUS system using the Oracle database, the ANJEZ system using the DB2 database, NOOR, ENTEMAA and ESTEBANA storing the data in the SQL database. As for the development of the thesis, it has started with a review of the state of the art of the topics of interest, namely: pattern mining in education, mining of academic curricula, personalization systems, etc. With this vision in mind, the general hypothesis has been put forward that knowledge extraction techniques can help obtain new methods that aid the development of decision support systems applied to LTCP. In this sense, two proposals for curriculum personalization have been developed: The first one is based on sequential pattern mining to discover which sequences of subjects are the ones that lead to successful profiles (better grades in the degree). This contribution has consisted of developing an evolutionary algorithm specially designed to analyse sequences of courses that differentiate students with excellent rates from the rest. Furthermore, these sequences establish difficulty indices to recommend subjects according to the previously taken issues. The approach yielded excellent results in the case study, producing interesting sequences for each student based on the previous courses they had already passed. The recommendation was based on the courses already taken by similar students, and an excellent final average was obtained. In addition, this methodology was able to provide complete study plans from the early stages of the course, which is essential for new students. A course difficulty index metric has been also proposed, and an online application is used to describe which courses are more difficult so that students can choose different courses according to a maximum difficulty value. The second proposal consisted of calculating a course difficulty index, called DMDIM, which can be used to carry out a proper estimation of the burden for students to choose a given set of subjects. The report shows the hypotheses that have led to the derivation of this index, illustrating it with several examples associated with KAU academic data. The results show that the index obtained is an excellent help in advising students on the load of subjects to choose from during an academic year. It also helps those responsible for educational organization design pathways to improve student performance. This second proposal has been implemented in a web application that automatically calculates all course difficulty indices, enabling teachers and/or students to obtain information on the teaching load of a given block of subjects to be chosen. In addition to the results achieved in the experiments, it should be noted that, as a general result, when designing departmental curricula, any educational institution should not only rely on credit hours or units but should also consider new factors such as the course difficulty index or the correct sequence of activities.