Predicting Academic Grades of Students in Computer Programming Using Classification Algorithms

Abstract

In computer science education, teaching and learning programming is difficult. Understanding and coding programmes are regarded as extremely difficult in computer science education. This is because practical ability is valued more than theoretical knowledge. According to research, students with metacognitive management skills outperform lower-performing students in programming. The more difficult the programming task, the more important it is for the programmer to have metacognitive control skills. The cognitive processes involved in learning computer programming necessitate the development of metacognitive skills in the novice programmer. This study’s main objective is to predict computer programming students’ academic grades using classification algorithms. The predictive analysis used 151 records that were gathered and used. Three methods are considered in order to find the best classifier: Rule Based Classification, Decision Tree Classification, and Nave Bayesian Classification. The confusion matrix, common metrics like precision, recall, ROC curve, kappa statistics, mean absolute error, root mean squared error, relative absolute error, and root relative squared error are all used to assess how well each classifier classified the dataset. Conclusion: With 95% accuracy, the ID3 classifier outperformed the other six predictive models.