Data mining is being increasingly leveraged in educational settings for achieving various different outcomes including students' learning patterns, course and teaching outcome assessment, and students' expected achievement prediction. Utilizing data collected from daily curricular and non-curricular activities, machine learning techniques have benefited administrators in making efficient decisions. Based on students' behavioral information, this research proposes student performance prediction model using fuzzy-based neural network (FNN) trained by a novel metaheuristic approach. Because original gradient-based learning method associated with FNN limits its performance, this research employs Henry Gas Solubility Optimization (HGSO) algorithm for tuning FNN parameters. The empirical analysis suggests superiority of results produced by the proposed approach as compared with the FNN trained by the competitive methods.
|Number of pages||9|
|Journal||Journal of Soft Computing and Data Mining|
|Publication status||Published - 4 Mar 2020|