Analysis and Comparison of Student Performance using Machine Language Algorithms

Abstract

In recent times, forecasting students’ Academic Performance has gained substantial traction and is the dynamic challenges of academic institutions. Educational Data Mining (EDM) with advanced techniques and methods plays a significant role in addressing students’ academic performance. Although research has been conducted at University and College levels, only limited research has been conducted at the School level, with respect to predicting academic performance.  Identifying students’ academic performances at an early stage is crucial for educational institutions and parents in order to take proactive decisions concerning a student’s future. The goal is to determine the factors that affect a student’s scholastic performance and enable parents and educational institutions to accurately predict a student’s academic performance and channelize the student’s capabilities in the right direction, at the right time.  This research makes use of Supervised Machine Learning approaches to analyze, filter and determine students at risk and suggest alternatives to improve their performance. In order to achieve the desired objective, this paper analyses Classification algorithms for EDM in depth, and identifies the right attributes and the most suitable Machine Learning tool which can accurately predict the scholastic performance of school students and ensure academic achievement. After conducting a comparative study of the results from various Machine Learning techniques using student data, this research shows that the Light Gradient Boost Method is  the best in predicting the scholastic performance of school students.