Data-Mining of the Centralized Test Scores of the University of Applied Science & Technology

The purpose of this study was to identify and discover unknown models and relationships and to identify cases outside the models, from the huge volume of data in the educational system of the University of Applied Science & Technology. For this purpose, the scores of 19207 students/courses in 8 courses, 6 provinces and 120 educational centers that have been centrally held throughout the country in the second semester of the academic year 1397-98 have been studied. Statistical analyses and data mining were performed using SPSS Modeler software. First, statistics were calculated and the relationship between different features and courses was examined. Then, by K-means clustering, 5 clusters were obtained. Cluster 1 is for poor students, who have low test scores. Cluster 2 is for excellent students who have a high number of credits and a high-test score. Cluster 4 is for excellent students who have taken fewer credits. These students have even higher test scores than cluster 2 students. Cluster 5 corresponds to the small number of credits taken, which includes conditional students. Students in this cluster have better test scores than weak students (cluster 1). Then the relationship between different variables was investigated. Among the variables, there is a direct relationship between test score and grade point average. The results of the research can be effective in advancing the educational goals of the university.