Forecasting Student Academic Performance Using Neuro-Fuzzy Model

Abstract

To achieve objective and precise forecasting of students’ academic performance while keeping up with technological improvements and changes, this paper discusses a neuro-fuzzy technique for the prediction of student academic performance. The technique comprises a Neural Network (NN), Fuzzy Logic (FL), Knowledge Base (KB) and Rule Base (RB) subcomponents. The NN subcomponent uses a three-layered feed-forward architecture to acquire a large collection of interconnected units in some pattern for communications. The FL subcomponent is an extension of Boolean logic and it was used for establishing accurate selection processes and precise solutions to multivariable problems. The KB component forms the database of multi-level information while the RB comprises a set of if-then statements for decision-making. The inference engine subcomponent applies a pre-defined procedure on input from the rule base and fuzzy logic interfaces for final prediction. The proposed technique performs predefined procedures that are based on some input sets that store multi-level information derived from several pre-specified scores. Results from the implementation of the technique established its practical function.