Efficiency of data mining models to predict academic performance and a cooperative learning model

Two purposes of this study are 1) to select a data mining model to predict learners' academic performance in computer programming subject to group learners for cooperative learning by comparing the efficiency of the models created from data mining with classification technique and 2) to develop a model for cooperative learning via web using the selected data mining model to group learners. The efficiency of seven models created from data mining with classification technique by using seven algorithms that are Artificial Neural Network, K-Nearest Neighbor, Naive Bayes, Bayesian Belief Network, JRIP, ID3 and C4.5 is compared and it was found that the models created from C4.5 has the best efficiency. The accuracy of the model created from C4.5 is about 74.8945% and the accuracy tests show that this model is reliable. Therefore this model is selected to group learners with STAD technique for cooperative learning through web. The result also shows that ID3 is inappropriate to predict learners' performance. The data mining model created from C4.5 shows that math's GPA has the most influential for academic performance in computer programming subject. The model for the cooperative learning model via web using C4.5 to group learners consists of 5 components that are data management module, prediction and grouping module, learning resources, cooperative community and quiz module. The results also show that in the case of using the selected model to group learners and in the case of grouping learners by the lecturers, the learning progressive-score in the first case is higher.