Analysis of Students' Performance in E-learning Based on Regression

Due to the spread of the Corona pandemic, e-learning has grown in popularity. Maintaining studenzt performance, participation and motivation are considered challenges were facing e-learning, especially the link between student participation and academic performance. This paper suggested using regression algorithms to predict students' performance and find the relationship between students' participation, the final exam result in the e-learning environment, and each effect's attribute on performance. To analyze students' performance and predict their performance, the Open University Learning Analytics (OULAD) dataset was used, which shows the students' interaction during the online laboratory work in terms of text editing, time spent on each activity, etc., in addition to the test score achieved in each session. Which are divided into three major groups depending on a variety of factors: (1) the kind of activity, (2) time statistics, and (3) the number of peripheral tasks. The suggested ML model forecasts whether a student will do poorly or well. Four popular prediction algorithmsXGboost, LASSO, Decision Trees, and Support Vector Regressionwere applied and tested in the investigation. The results showed that LASSO performed the best in terms of prediction accuracy