Comparative Analysis of Cluster and Random Tree Algorithm Based on Students Performance Outcome

Abstract

Student performance is the major tool use in determining the status of students at all levels of education.
Data mining tools are nowadays used in determining the students’ performance and it is greatly helps in
making analysis and decision based on the performance. This paper aims at comparing Cluster algorithm
and Random/Decision Tree algorithm. Rapid miner studio is used to determine the best algorithm to
determining student performance. We conducted this work in Mai Idris Alooma Polytechnic, Geidam Yobe
State, where student score sheets that contains six (6) attributes and 669 tuples was used as the dataset for
this work. One attribute was selected as a label attribute that determine the Student performance in case of
supervised learning, while on the other hand average within centroid distance of all the clusters is measured
to see closeness within performance of students. Student grade was used for determining performance of
students as label attribute. The findings show that Random Tree algorithm has a higher class precision with
an accuracy of about 73.73% compared to that of measures of average within centroid distance. The findings
will equally help in marking a sound academic planning in future. Finally, the analysis of the results
obtained will go a long way in making recommendations for future work.