Using Feed-forward Backprop, Perceptron, and Self-organizing Algorithms to Predict Students’ Online Behavior

Abstract

Pandemic situation has opened up an e-learning environment for students. Understanding of students’ reaction towards e-learning environment based on the evaluation of students’ performance to understand students’ behavior is very important. In the paper, techniques for evaluating the online reactions to predict behavior via students’ performance from their classmates are discussed. Data were collected about students from a Brazilian University and secondary education of two Portuguese schools for explorative data analysis. Feed-forward Back prop, Perceptron, and Self-organizing Algorithms using Matlab are applied to predict students’ behavior. The finding shows that the accuracy of Feed-forward Backprop, Perceptron, and Self-organizing algorithms is 68, 80, and 76 percent, respectively. The examination of students’ behavior is based on reactions from the assessment of learning outcomes and the usage of social features in the classroom.