Introduction
- Is it possible to predict the students’ grades using grades and attendance in subjects from past semesters? Does students’ attendance impact on their grades in the semesters? Research shows that attendance is one of the most important factor in students’ academic performance and achievement (Jones 2006; Kassarnig et al. 2017). Considering students’ attendance, we employed a variety of machine learning models to predict students’ data trends over several semesters and we compared this prediction to the real data, in order to measure their accuracy.
- Is there any correlation between students’ gender and their performance on different subjects? We considered gender in a performance statistical evaluation. As this factor as well as race, ethnicity, educational, and psychological factors are also addressed in some studies (Dee 2005; King et al. 2002; Wilson and Shrock 2001).
- Do external factors, such as parents’ education level impact on students’ performance in school? The influence of external factors is also discussed by Gooding (2001). We employed a “Multidimensional Projection" technique to explore the structure of the relationship among students in terms of similarities. The idea is to identify profiles and/or outliers that may explain their performance in the courses over the semesters.
Related work
Academic visual analysis system
Collection 1: exact science subjects performances
Attendance analysis & machine-learning approach
Attendance analysis & scatter plots
Gender analysis & statistical heatmap matrix
Gender analysis & course treemap
Logistic regression & pass/fail analysis
- TP: number of students correctly predicted to pass;
- TN: number of students correctly predicted to fail;
- FP: number of students incorrectly predicted to pass;
- FN: number of students incorrectly predicted to fail.