Prediction of secondary school students' success with machine learning

In today's world, artificial intelligence is utilized in numerous fields, and education is one of them. Alongside educational activities, machine learning methods can be employed to classify success, determine influential factors, identify causes of failure, and predict students' academic achievement levels in advance. It is crucial to anticipate this situation in order to take precautions and develop solutions in case of failure. If success situations can be predicted as accurately as failure, encouragement, and support can be provided to enhance and stabilize success. Twelve different research studies that use two distinct datasets prepared with secondary school students, along with various derivatives of these datasets were presented. In each study, eight machine learning algorithms were employed, namely: Deep Learning Algorithm, Artificial Neural Networks, Simple Logistic Regression, Iterative Classifier Optimizer, Decision Table, Fuzzy Rule Induction Algorithm, One-Rule and Logistic Model Tree. Initially, all attributes were employed to test classical, bagging, and boosting methods. Subsequently, a selection process was conducted to determine specific attributes for further analysis. In the studies conducted with feature selection, different algorithms were used to determine the features. The 10-fold crossover method was employed to predict success classes, and various performance metrics such as accuracy rate, precision, recall, and F-measure were calculated. Finally, the obtained results were compared with similar studies. As a result of the analysis, it was observed that there is a relationship between the characteristics of the dataset and the accuracy rate.