Ensemble learning-based prediction of contentment score using social multimedia in education

Over the years, social multimedia has gained credibility as a source of information and a reliable platform on which organizations, students and employees can interact with expert audiences. In the areas of education and teaching, the technique, multimedia and network of use computerized methods to build a creative environment for learners and broaden our perspective on a variety of topics. Getting new information and sharing it with others has become much easier with social multimedia. To boost the productivity and growth of any university, student contentment is a critical factor. Student contentment level is the need of the hour that is necessitated to be analyzed every year for the progress of the university. In this paper, we use a social multimedia technique to collect data from the students of the university based on a designed questionnaire circulated. The collected information embraces different aspects like academics, research, recreational, and technology that portray the image of the university. The current work relies on developing a stacking ensemble machine learning model for prediction of student’s overall contentment score, an indicator to perceive overall, how much the university gets the thumbs up from its current students. The work employs the cuckoo search meta-heuristic based wrapper method for feature selection from the original dataset with 78 features. The proposed ensemble model portrayed a lowest RMSE value of 0.373 by the combination of Self Organizing Map, Multilayer Perceptron, Boosted Generalized Linear Model and Gaussian Process with Polynomial Kernel along with Partial Least Squares as meta-learner, showcasing its ability to accurately predict student contentment levels of a University. The proposed machine learning framework acts as a great developmental tool for foreseeing and analyzing student contentment for its university.