Drop-out Identification model using Data Mining for an Intelligent Recommendation System for Universities in Thailand

In Thailand, many universities face the problems of student dropouts or failures before graduations. In order to improve and support the academic management processes, some universities are developing innovative information systems and services with an aim to enhance efficiency and retain the students to graduations. Moreover, this information technological support can also improve student relationship with universities. One of the key initiatives is the development of Student Relationship Management Systems (SRM) and among their functions, is the provision of recommendation and advice for students. Intelligent Recommendation systems allow personalization for counseling. The proposed system examined the correlation between 9000 student records and their academic results by analyzing student history data to associate with freshmen and current student data. In this experiment, Clustering followed with data mining; Artificial Neuron Networks (ANN), Support Vector Machines (SVM), Chi-squared Automatic Interaction Detector (CHAID) have been employed, the results of prediction techniques have been compared in order to choose the most accuracy. Then two aggregation models; Ensemble method and the Modular Artificial Neural Networks Optimized Weight of Subspace Reconstruction Method (MANN-OWSR) have been used to combine the results from the learning models for better performance. Good results have been obtained from the experiments and the developed system will help counselors, supervisors and academic staff in suggesting appropriate recommendations for the students.