Name
Abstract | ||
|---|---|---|
The main challenge in higher education is student retention. While many methods have been proposed to overcome this challenge, early and continuous feedback can be very effective. In this article, we propose a method for predicting student final grades in a course using only their performance data in the current semester. It assists students in analyzing how much effort they need to put into the course to obtain the desired grades while helping course instructors identify student types at the early stages of the course to provide better support for them. Our method, initially clusters students into several groups based on experience points (XP) that they obtain during a semester. Then, we estimate cluster size and balance clusters by generating and adding virtual students to the smaller clusters. Finally, we drop unimportant student attributes using a feature selection technique. We then predict their final grades via three different algorithms. We have compared the performance of our method with other approaches using data collected from a course for nine years, using data collected from 679 students. The results indicate that our method outperformed the others while achieving 78.02% average accuracy only four weeks after starting the course. It shows we can effectively predict final grades, which will potentially enhance students’ learning outcomes. |
Year | DOI | Venue |
|---|---|---|
2022 | 10.1109/TLT.2022.3170494 | IEEE Transactions on Learning Technologies |
Keywords | DocType | Volume |
E-learning,gamification,grade prediction,random forest (RF),virtual students | Journal | 15 |
Issue | ISSN | Citations |
3 | 1939-1382 | 0 |
PageRank | References | Authors |
0.34 | 27 | 4 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Amir Hossein Nabizadeh | 1 | 6 | 1.42 |
| Daniel Goncalves | 2 | 0 | 0.34 |
| Sandra Gama | 3 | 65 | 8.52 |
| Joaquim A. Jorge | 4 | 1008 | 81.51 |