Session 6B8 International Conference on Engineering Education August 6 – 10, 2001 Oslo, Norway 6B8-12 USING AN ACADEMIC DSS FOR STUDENT, COURSE AND PROGRAM ASSESSMENT Dervis Z. Deniz 1 and Ibrahim Ersan 2 1 Dervis Z. Deniz, Eastern Mediterranean University, Electrical & Electronic Eng. Dept., Gazimagusa, Mersin -10, Turkey, dervis.deniz@emu.edu.tr 2 Ibrahim Ersan, Eastern Mediterranean University, Electrical & Electronic Eng. Dept., Gazimagusa, Mersin -10, Turkey, ibrahim.ersan@emu.edu.tr Abstract  Decision-making necessitates accurate information to be available in a timely fashion. Too often information is not available to decision makers in a useful form or the available data has not been evaluated sufficiently to reveal hidden or crucial details. Universities need to have extensive analysis capabilities of student achievement levels in order to make appropriate academic decisions. Conversely, certain academic decisions will lead to changes in academic performance, necessitating periodic assessment for determining the effect of changes. In this work, a performance based academic decision support system employing data mining techniques is developed in order to extract useful information from raw data available in student databases. In addition, the different ways in which student performance data can be analyzed and presented for academic decision-making are studied. Application of the technique to student, course and program assessment is also investigated. Index Terms academic decision support system, assess- ment, data mining, performance statistics. INTRODUCTION Academic decision-making is a necessary part of university administration. Accurate and timely information is of paramount importance for informed decision making. Too often, the information is not available to decision makers in a useful form or the available data has not been evaluated sufficiently to reveal hidden or crucial details. Universities need to have extensive analysis capabilities of student achievement levels in order to make appropriate academic decisions. Conversely, certain academic decisions will lead to changes in academic performance, necessitating periodic assessment for the determination of the effect of changes. The study of academic decision-making process, effective resource management, personnel administration, automation of student registration, factors determining retention, graduation and dismiss have always been of great interest to educationalists. More recently, the issue of assessment has been a hot research topic. Various academic decision support systems (DSS) for academic planning, evaluation, advising and comparison have previously been proposed [1]-[7]. Kassicieh and Nowak [1] investigate the model-based DSS for allowing the decision-maker to plan and respond quickly to changes in academic environments. These include changes in admission policies, forecasts of student demand for particular courses, effects of introduction of different majors, program standards and modified GPA requirements, as well as some other well-defined changes. Turban et al. [2] review various approaches for use of DSS in academic administration. Modeling and simulation of student profile changes to determine retention and graduation rates using Markov chain analysis has been studied by Borden et al. [3]. Prediction and analysis of freshmen retention has been investigated within the context of cognitive, affective and psychomotor domains by Zhang and Richarde [4]. Bailey et al. [5] developed a model for predicting student retention. The use of a DSS for academic advising has been investigated by Murray and Le Blanc [6]. Performance evaluation of academic depart ments in a university in terms of productivity and quality has been the topic of research by Lopes et al. [7] for possible resource allocation amongst departments. Student performance data contain a wealth of hidden information. It is necessary to mine this information from raw data available in the student databases. An academic decision support system employing data mining techniques needs to be employed for consistent and standardized analysis of such data. Student performance data can yield valuable information about many related fields; the course and semester grade distributions for single or multiple terms, the regularity of students, regularity according to semesters, CGPA distributions at graduation, satisfactory/unsatisfactory status, time for program completion, retention levels, absent/dismissed status are but some examples. If student intake to engineering programs is from a wide-ranging background, students may experience different difficulties as may be indicated by the loss of regularity at different semesters or as a result of different courses. Identification of particular problems experienced by groups of students is helpful in determining bottlenecks in the program. This in turn may help to identify courses that must be given additional attention or in determining the adequacy of program admission requirements. Pre-requisite level of knowledge of students for success in certain courses can be better identified in this manner. In this work, the different ways in which student performance data can be analyzed and presented for academic decision-making are investigated. A software package called the Performance based Academic Decision Support System (PADSS) is designed and developed for this purpose. The software system makes extensive use of