Data-Driven Predictive Analysis of Student Performance In College Using Neural Networks

The scientific exploration of data for educational research is often referred to as Educational Data Mining (EDM). EDM concentrates upon devising methods for evaluating data coming from educational settings to understand students and the locale in which they study. This paper, in particular, encompasses those students who are currently pursuing their higher education. In spite of a substantial inclination of students towards getting a degree, the success rate is remarkably low. Numerous studies have been conducted, seeking to develop methodologies that identify students who are at risk of unsatisfactory performance. In our approach, we explore multiple factors that have been theoretically assumed to affect the performance of students in college and use neural networks to predict their grades. We also introduce the scientific assessment of course difficulty prior to using it as a measure for a students' performance in that course. The model can, therefore, be utilized to identify students who are most likely to perform under par and assist them in achieving better grades.