ANALYSIS OF STUDENT PERFORMANCE DATA: A COMPARISON OF ARTIFICIAL NEURAL NETWORKS and REGRESSION ANALYSIS

Abstract

The study was created with the effective variables to evaluate performance belongs to 649 students which ready in the ready data set in the database. Firstly, students were distinguished between successful and unsuccessful situations, and data entries were made as binary variables. Discriminant analysis was performed with the SPSS program in order to group variables that affect performance. The variables, whose distinctions were provided with the help of discriminant analysis, contributed to the models planned to be created. Artificial neural networks (ANN) method was preferred as the algorithm used in processing big data to predict student performances. In the application phase of the study, the two methods examined were analyzed with the help of the data set. The results obtained from the application were compared for 3 separate criteria: MSE-classification accuracy and the area under the ROC curve. At the end of the study, it was concluded that artificial neural networks produced more successful results than logistic regression