Applying data mining for predicting computer science students academic performance

Abstract

The aim of this thesis is to study the impact of a three-year-study of computer science at Notre Dame University on the students’ academic performances. For this reason, the GPAs of the students after their first year of study were compared with their GPAs upon graduation. To perform this, a Decision Tree, as well as Neural Networks were used. These algorithms helped us to predict the final GPA based on the first two semesters’ GPA. In addition, the most decisive courses on the student GPA were identified. For this reason, Relief algorithm was used to identify the three most important courses which affect the final GPA. This output helps instructors, managers, and even students to put more emphasis on these courses hoping for a better GPA. By using the F-measure, which is the weighted harmonic mean of the precision and the recall, the effectiveness of the algorithms is evaluated. The F-measure in this research for all the simulations performed ranged between 74.9% and 91.2%, which is statistically acceptable.