Knowledge Based Performance Evaluation and Predictive Model for Undergraduate Students

Abstract

In educational data mining, the process of analysing and predicting from a pool of acquired data is a big challenge due to the influence of behavioural, environmental, parental, personal and social traits of students. While existing education predictive systems have used patterns generated from mined common factors to predict student performance based on subject, faculty, and grade amongst others, explicit traits, which defines a student are often neglected. Thus, such existing models are too general for specific and targeted analysis in more recent times when predictive features are although common but in real essence unique to individual students to a certain degree. Here, a Self-Academic Appraisal and Performance Predictive (SAAPP) system was developed to analyse and predict the overall performance of students before the expiration of their course duration. The inherent knowledge driven model analyses common available predictive internal and external factors, with probabilistic analysis of student academic history and pending courses. The system then builds a personal data centric system for individual student through a decision support expert system and a probabilistic optimal grade point analysis for more effective recommendation. The developed system is more accurate, reliable and precise in student performance classification with targeted recommendations.