PASS: An Expert System with Certainty Factors for Predicting Student Success

In this paper, we present an expert system, called PASS (Predicting Ability of Students to Succeed), which is used to predict how certain is that a student of a specific type of high school in Greece will pass the national exams for entering a higher education institute. Prediction is made at two points. An initial prediction is made after the second year of studies and the final after the end of the first semester of the third (last) year of studies. Predictions are based on various types of student’s data. The aim is to use the predictions to provide suitable support to the students during their studies towards the national exams. PASS is a rule-based system that uses a type of certainty factors. We introduce a generalized parametric formula for combining the certainty factors of two rules with the same conclusion. The values of the parameters (weights) are determined via training, before the system is used. Experimental results show that PASS is comparable to Logistic Regression, a well-known statistical method.