Developing a model for predicting student performance on centralized test Based on Data Mining

The aim of this study is to provide a model for predicting University of Applied Science & Technology students' scores in centralized exams in the coming semesters of the university. For this purpose, the status of the 19/207 student/ course grades has been studied in 8 courses in 6 provinces and 28 educational centers, that have been held in an associate's and bachelor's degree level and concurrently across the country in the second semester of 1397-98 And by using the feature selection method, the most effective ones were selected. To clarify the relationships between the selected features and the decision tree model with C5.0 algorithm using SPSS Modeler software, with 10 effective indicators, a model for predicting students' scores in the next semester is presented in the courses approved for the centralized exam. This predictive model can be effective in making the learning process more efficient in the academic system. The results of these models include suggestions for modifying the test process, finding students and centers, and out-of-pattern conditions for further monitoring and identifying centers whose students' average GPAs were high but poor on the centralized test.