Analyzing Factors that Influence Student Performance in Academic

Abstract

Student performance analysis is a complex and popular study area in educational data mining. Multiple factors affect performance nonlinearly, making this topic more appealing to academics. The broad availability of education adds to this interest, particularly in online learning. Although previous studies have focused on analyzing and predicting students' performance based on their classroom activities, this study did not take into account student's outside conditions, such as sleep hours, extracurricular activities, and a sample of question papers that they had practiced. These three variables are included among others in our study. In this paper, we describe an analysis of 10,000 student records, each containing information on numerous predictors and a performance index. The dataset intends to shed light on the relationship between predictor variables and the performance indicator. To create the correlation variable heatmap, we use univariate and bivariate studies to produce a linear equation. The selection of these two techniques is based on their simplicity compared to other techniques. Besides, both are the most fundamental techniques in finding the data pattern. The bivariate analysis enables us to find the relation between two variables involved in the study. Finally, we showed the actual and expected student performance outcomes using the model we constructed. Following this, we perform data preprocessing and modeling to facilitate predictive analysis. The findings demonstrate that our prediction model was 98% accurate, with a mean absolute error of 1.62.