The use of educational data mining to predict student success

The aim of this study is to emphasize the importance of big data, learning analytics and data mining concepts in education and to conduct educational data mining studies to increase the academic performance of students. In this study, a logistic regression model was created to estimate the effects of university placement score, placement score type, high school score, gender, province, age and marital status variables on being able to graduate from university in 4 years. In addition to the variables mentioned above, it was investigated whether the variables of passing the Introduction to Computers I course and grade were related to the success in the Introduction to Computers II course and a logistic and linear regression model were also created. For this purpose, a data set of 223,279 students was created from the information registered in the student automation system of İnönü University. This data set was processed in accordance with the CRISP-DM business process steps, which is one of the educational data mining process designs. The editing of the data set and the creation of the models were carried out with the RapidMiner Studio program. As a result of the analysis, it was observed that the student graduation status logistic regression model performed at a high level with 76.80. It has been concluded that the graduation period can be predicted by using the personal and academic information of the students after their registration to the university, and the necessary measures can be taken for the success of the students by using these variables. It was seen that the logistic regression model of the Introduction to Computers II course showed a high level of performance with 79.34%, and it was concluded that whether the students will pass this course or not can be predicted at the beginning of the second semester by using their personal and academic information. The fact that the developed Introduction to Computers II course passing grade linear regression model can make low erroneous estimations made it possible to predict the grade that students will get from this course by using their personal (Gender, Age, Province) and academic (University Placement Score, Placement Score Type, High School Graduation Score, Introduction to Computers I Course Passing Grade) information. As a result of the research, it was emphasized that since the information stored in educational institutions is valuable, it is very important that it is meticulously stored and made accessible anonymously. It has been concluded that academic achievement can be increased by giving the necessary warnings and supports to the students in the early stages with the educational data mining prediction models made with smooth and large numbers of data.