Model proposal related to predicting student academic performance: A study based on data mining

In this study, an applied data mining study was conducted by using the academic data of 3773 students who graduated from Akdeniz University Faculty of Education between 2012-2017. Exam scores, semester grades of lessons, semester grade point averages and graduation grades data of the students from 7 departments of Akdeniz University Faculty of Education were gathered from different tables and separate data sets were obtained for each department. After the data pre-processing stage, academic performance prediction models were developed and tested. Two main aiming models to predict students' academic achievement were developed by using data mining techniques and algorithms. The first model was the Prediction Model of Student Graduation Grade. Under this model, three sub-models were developed as the 1S1DV Model for predicting the graduation grade by using midterm exam scores of the 1st semester, 1S1D4N Model for estimating the graduation grade through 1st-semester grade point averages and 1S12DANO Model for predicting graduation grades by using grade point averages of 1st and 2nd semesters. For each sub-model, different models were developed by using artificial neural networks and multiple linear regression analysis and their performances were compared. It was observed that the models, developed in this research, predicted the graduation grade of students with an accuracy of 94-97% since the first semester. The second model developed in this research was the Student Academic Early Warning (DANO2) Model. The DANO2 Model was a model that predicted whether the future grade point averages of the students would fall below 2 according to their 1st-semester grades. Under this model, the accuracy of the sub-models developed by using logistic regression and decision trees was 72-87%. As a result of this research, a model aiming to prevent future academic failures was proposed by predicting student academic performance. With this proposed model, it was thought that educational institutions could work more effectively and efficiently in increasing student success. Keywords: Prediction of Academic Performance, Academic Warning System, Educational Data Mining, Artificial Neural Networks, Decision Tree, Regression Analysis, Logistic Regression, Faculty of Education As a result of this research, a model aiming to prevent future academic failures was proposed by predicting student academic performance. With this proposed model, it was thought that educational institutions could work more effectively and efficiently in increasing student success. Keywords: Prediction of Academic Performance, Academic Warning System, Educational Data Mining, Artificial Neural Networks, Decision Tree, Regression Analysis, Logistic Regression, Faculty of Education