Modeling educational data in Afghanistan using machine learning methods

Abstract

This paper examines various factors affecting a student's academic performance at the stage of secondary education in Afghanistan. Of course, academic achievements play a crucial role in determining a student's future profession, as well as their goals in an academic or professional environment. Academic success is influenced not only by academic performance itself, but also by other factors, including individual differences. In this work, various machine learning algorithms were used to analyze the factors influencing a student's grades in the last grade of school, in particular, learning with a teacher (regression and classification) and without a teacher (clustering algorithms). The purpose of the study was to determine which factors significantly affect student performance for the entire sample under consideration, which factors are more individual in nature, as well as the degree of their influence. The results showed that various machine learning algorithms can be successfully used to determine the factors influencing students' academic success. Students' final grades are undoubtedly influenced by the results of their studies in previous years, as well as the time spent on learning. But it is very important to note that the family situation and the balance between study and leisure activities also affect the student's success. In particular, the level of education and the mother's job have a great impact on the student's academic success: if the student's mother is educated and, especially, if she works as a teacher, this significantly increases the student's chances of academic success.