A predictional model of educational success with the machine learning method

Today, machine learning methods have been used effectively and have shown high performance in many areas. It has become more widespread in various sectors in recent years. Many problems can be predicted and solved with the successes that can be obtained from machine learning models. The aim of this study is to present a machine learning model that will predict educational success with the data collected from the survey conducted with secondary school students and to present the factors that may affect the student. The questions of the questionnaire were created by investigating the factors that may affect the success of the student. Within the scope of the study, data were collected from 520 different students studying in various secondary schools through a questionnaire consisting of 13 questions within the scope of the law on the protection of personal data. This data is not shared with any institution and its confidentiality is protected. K-Nearest Neighbor(K-NN) , Random Forest (RF), Linear Regression, Bagged Trees, Gradient Boosting Regressor(GBM), Decision Trees (DT) algorithms were used. Among them, the Random Forest algorithm with the highest model success was selected. In the study, after the data manipulation processes were carried out, the model was established and the estimation of the educational success of the student was made based on the Turkish grade. Turkish course was chosen as the dependent variable. The determination of the course selection in the study is due to the fact that the mother tongue is Turkish and that Turkish course has been encountered every semester throughout the education life of a student.