International Journal of Man-Machine Studies - Special Issue: Knowledge Acquisition for Knowledge-based Systems. Part 5
C4.5: programs for machine learning
C4.5: programs for machine learning
Machine Learning
A Comparative Analysis of Methods for Pruning Decision Trees
IEEE Transactions on Pattern Analysis and Machine Intelligence
Machine Learning
On Estimating Probabilities in Tree Pruning
EWSL '91 Proceedings of the European Working Session on Machine Learning
Building a Recommender Agent for e-Learning Systems
ICCE '02 Proceedings of the International Conference on Computers in Education
Data Mining: Concepts and Techniques
Data Mining: Concepts and Techniques
Top 10 algorithms in data mining
Knowledge and Information Systems
Data Mining with Decision Trees: Theroy and Applications
Data Mining with Decision Trees: Theroy and Applications
Educational data mining: a review of the state of the art
IEEE Transactions on Systems, Man, and Cybernetics, Part C: Applications and Reviews
Review: Educational data mining: A survey and a data mining-based analysis of recent works
Expert Systems with Applications: An International Journal
Hi-index | 0.00 |
Student academic performance at universities is crucial for education management systems. Many actions and decisions are made based on it, specifically the enrollment process. During enrollment, students have to decide which courses to sign up for. This research presents the rationale behind the design of a recommender system to support the enrollment process using the students' academic performance record. To build this system, the CRISP-DM methodology was applied to data from students of the Computer Science Department at University of Lima, Perú. One of the main contributions of this work is the use of two synthetic attributes to improve the relevance of the recommendations made. The first attribute estimates the inherent difficulty of a given course. The second attribute, named potential, is a measure of the competence of a student for a given course based on the grades obtained in related courses. Data was mined using C4.5, KNN (K-nearest neighbor), Naïve Bayes, Bagging and Boosting, and a set of experiments was developed in order to determine the best algorithm for this application domain. Results indicate that Bagging is the best method regarding predictive accuracy. Based on these results, the "Student Performance Recommender System" (SPRS) was developed, including a learning engine. SPRS was tested with a sample group of 39 students during the enrollment process. Results showed that the system had a very good performance under real-life conditions.