The Design and Implementation of Intelligent Course Advisory System Using Learning Analytics

Abstract

The Students of class 12 in India choose a program from a pool of programs offered by higher newlineeducation institutions. As many undergraduate programs are introduced in an educational system, newlinechoosing an appropriate program is a challenge in the current era with an abundance of newlineinformation. Educational data mining, learning analytics and personalized recommender systems newlineare the widely used branches of computer science that facilitate the design of intelligent course newlineadvisory systems that are needed to help the students to select the appropriate undergraduate newlineprograms. newlineChoosing the right course in formative years is very important decision as his future depends on newlinethis one decision. The student by himself is not mature enough to take right decision in his early newlinelife. Selecting wrong courses means mismatch between student aptitude, capability, and personal newlineinterest. Faculty or parents have neither the required knowledge nor experience. Since there is no newlineother reliable source generally available that can guide the student towards the most suitable newlinedirection, the recommender system has been evolved to provide him guidance in selecting the right newlinecourse. newlineRecommender systems are widely used in different walks of life. This research work is to explore newlinethe usage of recommender systems in the field of education. As there are multiple program options newlinefor students after class 12, this research work is towards the recommendation of appropriate newlineprograms for the students by using collaborative filtering algorithms. newlineThe existing program selection strategy among the students is analyzed in the first step of the newlineresearch. A suitable framework to implement a recommender system is designed as a second step newlineof the research work. An extensive data collection using questionnaire is carried out and effective newlinepreprocessing techniques are executed to clean the data. In the next step, a simple recommender newlineengine is designed using an implicit rating matrix implementing collaborative filtering algorithm. newlineIn the next step, the context-aware recommender system is designed based on the predictive values newlinefor the various contextual parameters using a contextual modeling approach. A novel hybrid newlinerecommendation algorithm is implemented by using Cuckoo search, pulse coupled neural network newlineand convolutional neural network. The experiments between state-of-the-art methods and the proposed approach are conducted to newlineevaluate the performance of the proposed method. The results of the experiments are evaluated newlineusing the measures and metrics relevant to recommendation systems. Experimental results show newlinethat the proposed method outperforms state-of-the-art methods both in accuracy and efficiency. newlineThe detailed outcome of all the steps is elaborated in different chapters of the thesis. R, WEKA, newlineMS-Excel, MATLAB, and other libraries are used to implement this research work at different newlinestages. newlineAs of now, 53 undergraduate programs in different disciplines in Indian education are considered newlinefor implementation. As a future advancement, the performance of the system can be analyzed for newlinea greater number of courses in different disciplines like medical, paramedical courses. newlineRecommendations can be further enhanced by having hybrid recommender models. newline