A Novel Method for Higher Education Risk Prediction through Machine Learning Algorithms and Deployment of Sentiment Analysis Leveraging Azure Cognitive Services

Abstract

Predicting student dropouts is a highly challenging task in a Virtual Learning Environment (VLE). The flexibility of VLE is the learners can study at their own pace and this flexibility makes learners dropout of class easily. Early detection of at-risk students in HEI plays a significant role in avoiding dropouts. An efficient and reliable risk prediction model is necessary for the development of VLE. The primary aim of the research is to propose an early intervention system that is best suited for classifying at-risk students in cloud Virtual Learning Environment and to extract behavioral patterns of students by deploying the model in cloud platform. From previous studies it is evident that to predict student drop out is a difficult task. It is inexplicit that which algorithm is most excellent for predicting at-risk students in VLE, and what are the most appropriate features to be considered for various machine learning classifiers, as the determinants of attrition depend on multi-dimensional characteristics. During the time of registering for online classes, student data collected at the university is not sufficient for dropout prediction. In this scenario, our research is carried out in four stages that focuses on several factors that influence students academic performance in online classes through various cloud platforms newline