Weak Student Identification: How Technology Can Help
| Authors | |
|---|---|
| Year of publication | 2014 |
| Type | Article in Proceedings |
| Conference | Proceedings of the 13th European Conference on e-Learning - ECEL 2014 |
| MU Faculty or unit | |
| Citation | |
| Field | Informatics |
| Keywords | student performance; social network analysis; educational data mining; prediction; university information system |
| Description | Predicting students' academic performance has long been an important research topic in many academic disciplines. When students enroll in a course, a teacher usually does not know their knowledge and skills. The teacher is faced with a difficult situation how to estimate the students. The knowledge about students' performance can be used for assigning students to seminar groups with respect to their skills needed for the course. The reliable prediction would also help teachers to identify weak students in order to help them to achieve better grades. The main aim of this research is to clarify relationship between students' behavior and their performance at the beginning of the term when there are no data about students' attitude and motivation towards the course. We inspect different data mining techniques how to recognize weak and good students and how to validate designed methods. We also aim to develop a model of students' performance indicators. Results can be beneficially used by teachers to enhance the teaching process. The model will be utilized in the university information system accessible to teachers and the faculty management. The first part of the paper describes the university information system with the focus on the data stored in its database. The paper also contains the description of the preprocessing stage and feature selection algorithms in order to reveal the most significant attributes. The next section introduces the experiments that aim to identify students' performance. The experiments are based on a combination of data mining methods and social network analysis and are evaluated on the historical data originated from the information system. The final section concludes the paper with comparing the designed methods and emphasizing the best fitting approaches. We also describe the prediction model and discuss the results in detail. |
| Related projects: |