Prediction of Primary and Secondary Education Institutions Scholarship Examination success with artificial neural networks

The aim of this study is to reveal the extent to which the variables of the Primary and Secondary Education Institutions Scholarship Examination (İOKBS) predict the success of students in the scholarship exam with the help of artificial neural networks (ANN). In addition, in line with the findings obtained, it is aimed to contribute to the improvement of the quality and content of IOKBS and to review the variables used in the student selection process. The relational survey model and purposive sampling method were used in the study. The data sets of the study consisted of the written exam scores (independent variables) of 926 students who took the IOKBS at the 5th grade level in a provincial centre in the Aegean Region of Turkey in the 2023-2024 academic year, and the written exam scores (independent variables) of 603 of the same students in the 4th grade Turkish, Mathematics, Science, Social Studies, Religious Culture and Moral Knowledge courses. In addition, the IOKBS achievement status of these students (dependent variable) was included in the data set. In data analysis, ANN models were created using MATLAB 2023a classification interface. In the research, the highest possible performance was determined by Bayesian optimisation and five-fold cross-validation method was used. As a result, it was determined that the ANN model created to predict the 5th grade IOKBS achievement status from the 4th grade written exam scores had an accuracy rate of 87.7%; the ANN model using the 5th grade written exam scores had an accuracy rate of 91.1%; and the ANN model using the 4th and 5th grade written exam scores together had an accuracy rate of 91.7%. It was determined that ANN is an effective method in predicting the achievement status of the students with high accuracy rates.