Prediction of student performance in blended learning utilizing learning analytics data

Introduction. This paper is dedicated to the problem of predicting performance of students who use online platforms. The objective of this paper is to identify characteristics of how to predict students’ performance in blended learning based on learning analytics data. Materials and Methods. Primary methods that were used in the research: theoretical analysis and generalization of scientific research, theoretical and practical methods of pedagogical research, statistical processing of empirical data, machine learning and random events modelling. Results. Conducted research identified that predication has to be performed based on the criteria that define learning success. Metrics for that criteria can be obtained through learning analytics data. Classification of students into groups based on their performance has to be done every time students complete their assignments in order to timely identify low performers who require special attention from professors. To efficiently predict future performance we need to accumulate dynamics of how students get re-classified into groups using discrete Markov Chains. Conclusions. Prediction of students performance based on learning analytics data allows identification of students who fall into high risk group, prediction of how students will be distributed among performance groups, and if necessary correction of teaching material.