Performance Evaluation of a Student Web Portal Using Classification Models - Data Mining Technique

Olanrewaju S. S.(1), Osunade O.(2), Ayeni J. A.(3), Amoo J. O.(4),

(1) Department of Science Education, School of General Studies Education, Federal College of Education (Special), Oyo.
(2) Department of Computer Science, Faculty of Science, University of Ibadan.
(3) Department of Computer Sciences, Faculty of Natural Sciences, Ajayi Crowther University
(4) Department of Computer Science. School of Secondary Education Science Program, Federal College of Education (Special), Oyo
Corresponding Author


Data Mining is an extraction tool for analyzing and retrieving hidden predictive information from a large amount of dataset. It has been discovered that huge amounts of data were automatically saved by the web portal, which contains hidden information about the client that assessed the portal and lots of these data remain unused. To get required hidden information from such large data, a powerful tool known as weblog expert analyzer version 9.51 was used to determine the performance of a student web portal. This research aims at determining the performance evaluation of a student web portal using weblog analyzer and data mining technique to predict the algorithm that gives the best accuracy in terms of model building. In this work, the machine learning software (collection of machine learning algorithms) called WEKA was used and the training parameter was set to 10-fold cross-validation. Five decision algorithms which are Rep tree, Random tree, J48, LMT, and Hoeffding algorithms were used as classifiers, while TP, FP, Recall, Precision, and F-measure, were used as evaluation metrics. Therefore, the Random tree algorithm yielded a 99.0138% higher level of predictive accuracy and provided better classification accuracy when compared to the other classifiers. Finally, the results obtained can be used to enhance the effectiveness of the student web portal.


Performance Analysis, Web Portal, Data Mining Technique, Predictive, Classifiers.

Full Text: PDF

Article Metrics

Abstract View : 177 times
PDF Download : 48 times


  • There are currently no refbacks.