To search, Click below search items.

 

All Published Papers Search Service

Title

Classification, Visualization and Pattern Recognition using J48 and Zero R Machine Learning Algorithms

Author

Ezekiel U Okike, Merapelo Mogorosi

Citation

Vol. 20  No. 12  pp. 224-231

Abstract

The identification and recognition of patterns is vital in Data mining and knowledge discovery. In the educational environment, mining data from Learning Management Systems (LMSs) data sets for useful strategic decision making has seen little investigation. Data was mined from Moodle LMS using the WEKA tool. J48 and Zero R ML algorithms selected from the WEKA tool were used to cluster, classify and visualize the data. The patterns of teaching and learning suggested that for clustering, both algorithms rated Quiz as the most used resource on the LMS resources followed by system logs which indicated that staff and students log on to the system to use the resources. In terms of classification, J48 is a better classifier than Zero R at 0.0, and 0.1 mean absolute errors, respectively. However, in terms of visualization of patterns, Zero R uses concentric colour coding while J48 uses tree format which becomes complex with large data sets. Therefore, Zero R is considered a better visualizer than J48. Overall, it was observed that there is significant correlation between students’ use of LMS resources and academic performance in the sampled test case.

Keywords

Data mining, Machine learning, Learning management systems, J48 algorithm, Zero R algorithms.

URL

http://paper.ijcsns.org/07_book/202012/20201225.pdf