Academic performance of recently accepted students is one of the main issues in Higher Level Institutions since first scholar periods trend to be the most difficult ones for students. Some institutions offer leveling courses to develop students basic knowledge for later courses. However, it is not clear if these help students in more advanced courses. This work presents an analysis, using decision trees, for predicting marks in two mathematics courses based on different criteria of the performance on a previous leveling course. This allows finding the factors that impact in the marks obtained in posterior courses and determining if the leveling one is helping students to improve their academic performance.
This work presents an evaluation of the predictive techniques decision trees using CART algorithm, Naïve Bayes Classifier, Gradient Boosting Machine and Support Vector Machine for predicting whether a student will successfully complete a programming course or not. Factors considered for prediction were university-entrance and personal criteria like entrance age, gender, scholarship, high school GPA, mark in admission exam and other related with student's performance in a prerequisite introductory programming course. The predicted variable takes two values, 'Approved' or 'Not Approved', and the data record contains an unbalanced portion of the class 'Approved'. For the analysis were considered two data sets, unbalanced and balanced. Evaluation of algorithms was performed considering the concepts of accuracy and ROC area. Results show that accuracy is bigger for the unbalanced data set, but its ROC area was very poor. Using the balanced data set, results were more reliable because accuracy and ROC area are closer. Best results were obtained with Naïve Bayes and Support Vector Machine algorithms. The most important factor in the prediction was whether a student had a scholarship or not.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.