International Journal of Computer and Information Technology (ISSN: 2279 – 0764) Volume 02– Issue 04, July 2013 www.ijcit.com 648 Framework of Recommendation System for Tertiary Institution Kuyoro ‘Shade O., Awodele Oludele, Okolie Samuel O. Department of Computer Science Babcock University Nigeria Goga Nicolae University of Groningen, The Netherlands or Politenica Bucharest Abstract— Understanding the reasoning behind variation in student academic performance in tertiary institutions has been the concern of many researchers for decades. Numerous studies have used traditional statistical methods to identify factors that affect and predict student performance. Machine learning has been successfully applied to so many domains, thus recently, researchers are employing this paradigm for modeling student academic performance and other related problems in higher education. This work focus at addressing the following: proposing optimal algorithm suitable for predicting students academic performance; designing a framework of intelligent recommender system that can predict students’ performance as well as recommend necessary actions to be taken to aid the students and identifying background factors that affect students’ academic performance in tertiary institution at the end of first year. This research used ten classification models and a multilayer perceptron -an artificial neural network function- generated using Waikato Environment for Knowledge Analysis (WEKA). Each model was built in two different ways: the first was built using the 10-fold cross validation, and the second using holdout method (66% of the data was used as training and the remaining as test). Purposive and selective sampling techniques were used in selecting one thousand five hundred (1,500) enrolment records of students admitted into computer science programme Babcock University between 2001and 2010. Results of the classifiers were compared using accuracy level, confusion matrices and speed of model building benchmarks. The random tree identified as optimal in this work is incorporated into designing a framework of intelligent recommender system. The work shows that identifying the relevant student background factors can be incorporated to design a framework that can serve as valuable tool in predicting student performance as well as recommend the necessary intervention strategies to adopt. Keywords- decision trees, neural networks, background factors, educational planning activities, machine learning, intelligent recommender system. I. INTRODUCTION Higher education systems all over the world nowadays are challenged by the new information and communication technologies.[1] With the increasing competition among higher education institutes, most are focusing on how to increase student retention rates and number of completions. University performance is one of the means of measuring its quality and reputation [2]; thus higher education institutions are becoming more interested in predicting the paths of students, and identifying which students will require assistance in order to graduate. [3] Higher learning institutions encounter many problems which keep them away from achieving this objective. Some of these problems stem from knowledge gap. Knowledge gap is the lack of significant knowledge at the educational main processes such as counseling, planning, registration, evaluation and marketing. For instance, many learning institutions do not have access to the necessary information to counsel students, thus they are unable to give suitable recommendation to the students. Also, there is a growing interest and concern in many countries about the problem of school failure and the determination of its main contributing factors. This problem has been referred to as “the one hundred factors problem”. [4]. Studies have shown that not all students who enroll as freshmen will complete their studies. [5]. The study of factors that influence the academic performance of students in higher education has a long history. The philosophy behind these studies is to understand what is related to, or predicts, poor academic performance and to use this information to design appropriate interventions [6]. A plethora of factors have been found to predict or influence retention and performance, falling into a number of broad categories: including gender, personality factors, intelligence and aptitude tests, academic achievement, previous college achievements, and demographic data, student background and demographics, prior educational achievement and level, psychosocial factors and approaches to study, and institutional and course factors.[7][8] However, factors found to be predictive in some studies are not always predictive in others [9], due in part to the ways in which different studies are designed. Indeed, even in the same study with the same methodology, results for student cohorts sampled at different universities have differed [10] and in general the results of particular studies cannot be generalised to other environments [7]. Machine learning is considered the most suitable technology in giving additional insight into educational entities such as; student, lecturer, staff, alumni and managerial behavior. It acts as an active automated assistant in helping them make better decisions on their educational activities. So far, there are limited numbers of research that examine the influence of family factors on first