Modeling Educational Data With Machine Learning Methods

Abstract

In our country, the effect of the academic success of the student, especially in the secondary education period, on the stage of choosing the profession he will have in the future and on the academic career goal is an undeniable reality. Academic success is affected not only by the data belonging to the academy, but also by many different categories. It is affected by many factors, especially methodological, and this diversity increases with individual differences. Regression and Classification from supervised learning models and Clustering algorithms from unsupervised learning models were applied to the data set. Multiple linear regression, polynomial regression, Lasso and Ridge regressions,Decision Tree, Random Forest, Support Vector Regression as regression methods, Decision Tree, Random Forest, Support Vector Machine, Logistic regression, K Nearest Neighbors methods were used as classification methods. As Clustering methods we are used K means algorithms, hierarchical method as unsupervised learning methods. In addition Artifical Neural Network, a deep learning algorithm, were applied to the data set. In the study, these factors and sub-factors were evaluated categorically and machine learning was used. Various determinations were made with estimation algorithms by establishing relations that predict the academic achievement target variable . By evaluating the data results, it is aimed to determine which factors affecting success are significant according to the sample group studied, which variables affect success individually and categorically, and the degree of influence, and as a result, it is aimed to contribute to education.