Analysis of student performance using Machine learning Algorithms

Abstract

Education is a prerequisite for a prosperous and good life, and it also helps in enhancing people's lives with meaning and excellence. Furthermore, education is viewed as a fundamental prerequisite for building self-confidence and providing the resources required to participate in today's speedily changing world. The progress of the educational institute's students can be used to quantify the institute's growth. Furthermore, education is viewed as a fundamental prerequisite for building self-confidence and providing the resources required to participate in today's rapidly changing world. For academic institutions and educators, analyzing student academic performance is critical in order to determine how to improve individual student performance. Using machine learning (ML) algorithms, this paper introduces a paradigm for forecasting students' academic success. This project examines past student outcomes, as well as their individual characteristics such as family history, demographic distribution, age, study attitude, and put this information to the test using diverse machine learning (ML) algorithms in WEKA (Waikato Setting for Knowledge Analysis) tool. The performance of the various algorithms was assessed using the percentage split (80:20) as well as the test-case cross-validation(10-fold). The results show that Linear regression (LR) is the utmost effective algorithm for forecasting student success, with a mean absolute error of 0.803 using cross-validation, and Artificial Neural Networks (ANN) is the least efficient, with a mean absolute error of 1.183 using percentage split.