Relevance and Resonance of Data Science in Performance Prediction and Visualization

Objectives: This paper aims at comparing and contrasting relevant researches on predicting the performance of students in Data Science perspective that encompasses machine learning and data mining. Analysis: An architectural framework has been devised for educational data mining. The ultimate motive is to extensively investigate the techniques like classification, regression and recommender systems in predicting the student performance and to explore the prediction accuracy of these techniques as well. For this purpose very many researches that have successfully implemented these techniques were carefully studied and their contribution to predicting the performance were analysed. Findings and Novelty: It came to light that ensembles created by combining classifiers performed well and their accuracy in predicting the performance was commendable compared to the individual performance of the classification, regression and recommender techniques. The nuance of this study is the incorporation of recommender systems along with conventional techniques since these are not commonly used in performance prediction. Tensor factorization in particular has desirable effect in prediction since it takes the time factor into consideration. It is a fact that performance of students increases over time.