New Classification Algorithms for Developing Online Program Recommendation Systems
Thomas Meller, Eric Zengxiang Wang, Fuhua Oscar Lin, Chunsheng Yang · 2009
This paper presents two novel nearest-neighbor-like classification algorithms for program recommendation in a Web-based system, which provides a program planning service to academic advisors and students of post-secondary institutions. To evaluate the accuracy of classification for program recommendations generated by our algorithm, a statistical study was conducted through comparing our algorithm against two well-known classification algorithms, the Naïve Bayes algorithm and the J48 algorithm, for making recommendations to students based on their academic history. The study shows that our proposed nearest-neighbor-like algorithms outperform the two well-known classification algorithms in terms of student classification success rate when there is uncertainty present in the data.