On the Design of a DSS for Academic Achievement Prediction

This paper tries to examine the relationship between students’ overall academic performance (GPA),
students’ grade of each subject of the first semester and their the high school grade, then comparing
the obtained results to highlight which is more likely to be predicted from the high school grade,
would it be the GPA or the grade of each subject by itself. This is done using the Decision Trees
algorithm for predicting the academic performance of the first semester for the undergraduate
engineering students at the Modern Academy for Engineering (MAE) by using the high school grade
as the only input. The data-mining tools were able to achieve levels of accuracy for predicting student
performance:
Decision Trees score for the {pass, fail} set scored 72% for the “Mechanics” which was the least one
while the highest score was for “Chemistry” with a score 89%, and as for the GPA grade the score
was 80%. For {excellent, very good, good, pass, fail, very bad, absent} set, the score was much less
for all of them and had a wide range of variance, it reached a minimum of 34% for “Physics” and a
maximum of 62% for “English” while the GPA grade scored 42%.
In this analysis, the Decision Tree was more accurate predicting at the {pass, fail} than at the
{excellent, very good, good, pass, fail, very bad, absent} data sets. The results of these case studie
give insight into techniques for accurately predicting student performance.