Performance Evaluation of Computer Science Students in Mathematical Courses Over Prgramming Language Courses

Abstract

Abstract: The low level of performance observed among many computer science students in mathematical and computer programming courses served as the motivation for this work. The prediction of computer science students’ overall performance in mathematical courses over programming courses was done using WEKA and machine learning algorithms. Dataset of students performance from Yaba College of Technology, Lagos, Nigeria and Lagos State Polytechnic, Lagos, Nigeria, for both National Diploma and Higher National Diploma were used. These data set were studied and analyzed using WEKA and Random Forest, J48, Naïve Bayes and Logistic Regression algorithms. The algorithms were applied for the HND and ND dataset and there was comparison based on their accurancy, learning time and percentage of correctly classified instances. This comparison showed that there is direct relation between the execution time in building models and volume of data records. This shows that the predictor did not only predict the number of students that are likely to be in distinction, upper credit, lower credit, pass and fail but also show the relationship between having the knowledge of mathematics and programming language for an overall performance in computer science. The knowledge pattern represented further satisfies the exertion that it is imperative for students to have a standard knowledge of mathematics as this will help in being the best in their chosen profession.