BINAR CLASSIFICATION OF STUDENTS' PSYCHOLOGICAL CONDITION USING XGBOOST

Abstract Nowadays, machine learning algorithms are used to solve issues in each sphere, including the field of education. In this article, we will consider a classification model based on the XGBoost (Extreme Gradient Boosting) machine learning algorithm to classify the psychological condition of schoolchildren into “stressed” and “non-stressed” classes. Currently, there is an increase in the number of factors that negatively affect the psychological condition of students, so this issue is important. Because excessive stress prevents students from receiving effective education. A dataset was created based on the psychological characteristics of schoolchildren for binary classification. A web application was created for the issue considered in the article. The XGBoost regularization of the binary classification in the article effectively solved the issue of overfitting due to the optimization of gradients and parallel computing capabilities. 326 students at school took psychological test via the web application. We will discuss the results in the paper.